A  B  C  D  E  F  G  H  I  J  K  L  M  N  O  P  Q  R  S  T  U  V  W  X  Y  Z  

  Subjects -> ELECTRONICS (Total: 207 journals)
The end of the list has been reached or no journals were found for your choice.
Similar Journals
Journal Cover
Selected Topics in Applied Earth Observations and Remote Sensing, IEEE Journal of
Journal Prestige (SJR): 1.547
Citation Impact (citeScore): 4
Number of Followers: 66  
 
  Hybrid Journal Hybrid journal (It can contain Open Access articles)
ISSN (Print) 1939-1404
Published by IEEE Homepage  [228 journals]
  • Snow Drought Patterns and Their Spatiotemporal Heterogeneity in China

    • Free pre-print version: Loading...

      Authors: Yuxin Li;Xiaodong Huang;Ying Ma;Qisheng Feng;Tiangang Liang;
      Pages: 2029 - 2036
      Abstract: This study presents a snow drought assessment method using snow water equivalent products to examine the patterns and differences in snow drought events in China from 1980 to 2020. The findings indicate that snow drought changes over the past 40 years can be categorized into three stages: The most severe snow drought occurred in the 1980s, followed by alleviation until 2009, and a subsequent aggravation after 2010. Light snow drought has the widest distribution and shows an increasing trend, whereas medium, heavy, and extreme snow droughts decrease gradually but show a decrease followed by an increasing trend over time. The distribution of snow drought in China displays significant spatial heterogeneity, with areas such as Alxa League in Inner Mongolia, Hami, Turpan, Bayingol, Xinjiang, and the Tibetan Plateau hinterland experiencing frequent snow drought events. Moreover, the low-altitude region has the largest average annual proportion of extreme drought, whereas the high-altitude region has the highest proportion of heavy drought. Among the three snow-dominant areas, Northern Xinjiang has the largest proportion of snow drought areas. The Northeast-Inner Mongolia has the highest proportion of extreme drought while the Tibetan Plateau exhibited abrupt changes in snow drought occurrence. These results contribute to a better understanding of the underlying mechanisms of snow drought changes, as well as serve as a foundation for the protection of the ecological environment.
      PubDate: MON, 01 JAN 2024 09:18:09 -04
      Issue No: Vol. 17, No. null (2024)
       
  • On the Correlation Between Earthquakes and Prior Ionospheric
           Scintillations Over the Ocean: A Study Using GNSS-R Data Between 2017 and
           2021

    • Free pre-print version: Loading...

      Authors: Badr-Eddine Boudriki Semlali;Carlos Molina;Hyuk Park;Adriano Camps;
      Pages: 2640 - 2654
      Abstract: From 1980 to 2021, earthquakes have caused more than 846 000 casualties and about US$ 661 billion in economic losses. At present, there are no reliable earthquake precursors to generate alerts. Currently, the link between earthquakes and total electron content variations measured by global navigation satellite systems (GNSSs) monitoring ground stations has been studied. However, GNSS ground-based monitoring stations are irregularly disseminated around the globe with significant gaps, particularly in the ocean's regions. In this article, we analyze ionospheric intensity scintillation anomalies as potential proxies of earthquakes. NASA CYGNSS GNSS-R (reflectometry) acquired by CYGNSS/delay Doppler mapping instrument from 2017 to 2021 has been used to calculate and analyze the anomalies in the S4 parameter over ocean areas affected by earthquakes. More than 30 000 ocean earthquakes within ±40° in latitude and with a magnitude larger than M4 have been examined. The daily planetary geomagnetic index Kp was used to discard data that may be disturbed due to space weather conditions. In addition, a daily sea wind speed mask was used to eliminate sea states that impact the signal reflectivity. The standard deviation and the interquartile time series methods have been used to detect these S4 anomalies. The confusion matrix, the receiver operating curve, and some other figures of merit have been used for the first time to evaluate and improve the performance of the prediction parameters and identify the optimum configuration to be used as a potential proxy of earthquake occurrence. As a result, a small, but detectable positive S4 anomaly was detected between 1 and 7 days before the earthquakes under study.
      PubDate: MON, 01 JAN 2024 09:18:09 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Adaptive Self-Paced Collaborative and 3-D Adversarial Multitask Network
           for Semantic Change Detection Using Zhuhai-1 Orbita Hyperspectral Remote
           Sensing Imagery

    • Free pre-print version: Loading...

      Authors: Dawei Wen;Xin Huang;Qiquan Yang;Jianqin Tang;
      Pages: 2777 - 2788
      Abstract: In recent years, numerous change detection methodologies have been proposed, with a predominant focus on binary change detection. Furthermore, there exists a paucity of research addressing semantic change detection in scenarios where solely binary change labels are available. This article introduces a multitask network for semantic change detection. First, 3-D ResUnet model is employed to generate initial multitemporal land cover results through postclassification comparison. Subsequently, the multitask network, encompassing two subtasks—binary change detection and multitemporal semantic segmentation—is proposed. Specifically, the shared branch of the network employs 3-D residual blocks to extract joint spectral-spatial features. In the subsequent task-specific branch, a 3-D GAN is incorporated for the binary change detection task to enhance the discrimination ability of latent high-level features for changes. Novel adaptive self-paced learning and certainty-weighted focal loss are proposed for multitemporal semantic segmentation to mitigate adverse effects from noisy semantic labels by considering sample complexity and reliability in the network optimization process. Experiments conducted on the Orbita Hyperspectral dataset in the Xiong'an New Area demonstrate the superior performance of the proposed method, achieving 99.28% and 76.60% for overall accuracy and kappa, respectively. This outperformance is notable when compared to other methods, such as Str4 and Bi-SRNet, showing an increase of 39.82% and 54.17% for kappa. Moreover, comparative experiments on SECOND further confirm the advantage of the proposed method, achieving 54.62% for kappa and outperforming other comparative methods, such as Bi-SRNet (47.61%).
      PubDate: MON, 01 JAN 2024 09:18:11 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Efficient Management and Processing of Massive InSAR Images Using an
           HPC-Based Cloud Platform

    • Free pre-print version: Loading...

      Authors: Zherong Wu;Peifeng Ma;Xinyang Zhang;Guangen Ye;
      Pages: 2866 - 2876
      Abstract: Significant progress has occurred in interferometric synthetic aperture radar (InSAR), emerging as a crucial technique for monitoring surface deformation. This evolution is attributed to expanded synthetic aperture radar (SAR) data availability and improved data quality. However, effectively managing and processing SAR big data presents substantial challenges for algorithms and pipelines, especially in large-scale contexts. In this article, we introduce a parallel time-series InSAR processing platform that leverages high-performance computing (HPC) clusters for efficiently managing and processing large-scale SAR data and incorporates graphics processing unit (GPU) acceleration to significantly enhance the speed and efficiency of specific InSAR processing algorithms. Our approach encompasses high-quality data compression, integration of classic InSAR models, and the introduction of a robust distributed scatterer InSAR method for time-series processing. The platform efficiently handles massive data, featuring a parallel optimization tool for acceleration. In addition, it provides web-based two-dimensional (2-D) result visualization and 3-D outcome representation for comprehensive user understanding. To illustrate our platform's capabilities, we applied it to 40 Sentinel-1 SAR data scenes from Tibet (2017–2019). Our data compression technique notably reduces data size, reducing mask data by 87.5% and coherence data to 25% of its original size. Leveraging HPC and GPU, we achieved a 50% reduction in registration computation time. This study offers valuable insights and a comprehensive platform for InSAR practitioners, facilitating calculations and enhancing comprehension of surface deformation processes. Our system's improved processing efficiency, coupled with a variety of InSAR methods, makes it an alternative choice for InSAR data handling and analysis.
      PubDate: FRI, 05 JAN 2024 09:18:18 -04
      Issue No: Vol. 17, No. null (2024)
       
  • HODet: A New Detector for Arbitrary-Oriented Rectangular Object in Optical
           Remote Sensing Imagery

    • Free pre-print version: Loading...

      Authors: Lu Xu;Jiawei Yu;Dongping Ming;
      Pages: 2918 - 2926
      Abstract: Object detection from remote sensing images is a key technology for Earth observation applications, which has important scientific research value. Ground objects in remote sensing images appear at arbitrary angles. However, object detection based on horizontal bounding boxes (HBBs) would cause mutual coverage among targets, while the ground objects were densely distributed or with a large aspect ratio. The oriented object detection methods could solve the problem by predicting the rotation angle. But the currently used methods were time-consuming in labeling and require complex loss functions of networks. Thus, this article combined HBB and oriented object detection to achieve rectangular object detection. First, an object detection network with HBBs is trained and the model predicts results in the original remote sensing image. Second, the rotation angles are derived from the line detection with linear Hough transform on the cropped region of targets obtained by the first step. Then, the original image is rotated according to the rotation angles and detects the object with HBBs again. Finally, new bounding boxes are mapped to the original image to get the detection results with oriented bounding boxes (OBBs). Experiments on public remote sensing datasets demonstrate that the proposed method is highly flexible and can be combined with any HBBs object detection network. The idea of original image rotation does not use the OBB labels to retrain the network, which reduces the workload of object annotations. In addition, the method can be used to automatically generate OBBs to label rectangular objects since the public datasets are commonly annotated by HBBs.
      PubDate: THU, 04 JAN 2024 09:17:23 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Flood Inundation Extraction and its Impact on Ground Subsidence Using
           Sentinel-1 Data: A Case Study of the “7.20” Rainstorm Event in Henan
           Province, China

    • Free pre-print version: Loading...

      Authors: Qianye Lan;Jie Dong;Shangjing Lai;Nan Wang;Lu Zhang;Mingsheng Liao;
      Pages: 2927 - 2938
      Abstract: On July 20, 2021, the northern Henan Province was hit by a historically rare, exceptionally heavy rainstorm (“7.20” Rainstorm Event), accompanied by severe urban flooding, flash floods, landslides, and other multiple disasters, resulting in significant casualties and property losses. On the other hand, the long-term overexploitation of groundwater since the last century has led to severe ground subsidence in the same area. We apply the intensity information of Sentinel-1 SAR images to extract the large-scale flood inundation area and their phase information to measure the ground deformation. Since heavy precipitations can recharge groundwater, the relationship between flood inundation, groundwater level change, and ground deformation is analyzed. The results show that the flood inundation areas are mainly distributed along the major rivers due to river overflowing, while heavy precipitation led to the rise of groundwater levels, and there was a significant amount of subsidence mitigation and surface uplift across the region due to the groundwater recovery. This study demonstrates the contribution of radar remote sensing to analyze the mechanism of groundwater recharge and subsidence mitigation benefited by rainstorm events and provides a technical reference to similar circumstances.
      PubDate: MON, 01 JAN 2024 09:18:09 -04
      Issue No: Vol. 17, No. null (2024)
       
  • A Novel Bottleneck Residual and Self-Attention Fusion-Assisted
           Architecture for Land Use Recognition in Remote Sensing Images

    • Free pre-print version: Loading...

      Authors: Ameer Hamza;Muhammad Attique Khan;Shams ur Rehman;Mohammed Al-Khalidi;Ahmed Ibrahim Alzahrani;Nasser Alalwan;Anum Masood;
      Pages: 2995 - 3009
      Abstract: The massive yearly population growth is causing hazards to spread swiftly around the world and have a detrimental impact on both human life and the world economy. By ensuring early prediction accuracy, remote sensing enters the scene to safeguard the globe against weather-related threats and natural disasters. Convolutional neural networks (CNNs), which are a reflection of deep learning, have been used more recently to reliably identify land use in remote sensing images. This work proposes a novel bottleneck residual and self-attention fusion-assisted architecture for land use recognition from remote sensing images. First, we proposed using the fast neural approach to generate cloud-effect satellite images. In neural style, we proposed a 5-layered residual block CNN to estimate the loss of neural-style images. After that, we proposed two novel architectures, named 3-layered bottleneck CNN architecture and 3-layered bottleneck self-attention CNN architecture, for the classification of land use images. Training has been conducted on both proposed and original neural-style generated datasets for both architectures. Subsequently, features are extracted from the deep layers and merged employing an innovative serial approach based on weighted entropy. By removing redundant and superfluous data, a novel chimp optimization technique is applied to the fused features in order to further refine them. In conclusion, selected features are classified using the help of neural network classifiers. The experimental procedure yielded respective accuracy rates of 99.0% and 99.4% when applied to both datasets. When evaluated in comparison to state-of-the-art methods, the outcomes generated by the proposed framework demonstrated enhanced precision and accuracy.
      PubDate: MON, 01 JAN 2024 09:18:11 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Three-Dimensional Modeling and Visualization of Single Tree LiDAR Point
           Cloud Using Matrixial Form

    • Free pre-print version: Loading...

      Authors: Fayez Tarsha Kurdi;Elżbieta Lewandowicz;Jie Shan;Zahra Gharineiat;
      Pages: 3010 - 3022
      Abstract: Tree modeling and visualization still represent a challenge in the light detecting and ranging area. Starting from the segmented tree point clouds, this article presents an innovative tree modeling and visualization approach. The algorithm simulates the tree point cloud by a rotating surface. Three matrices, X, Y, and Z, are calculated by considering the middle of the projected tree point cloud on the horizontal plane. This mathematical form not only allows tree modeling and visualization but also permits the calculation of geometric characteristics and parameters of the tree. The superimposition of the tree point cloud over the constructed model confirms its high accuracy where all the points of the tree cloud are within the constructed model. The tests with multiple single trees demonstrate an overall average fit between 0.3 and 0.89 m. The built tree models are also compliant with the Open Geospatial Consortium CityGML standards at the level of a physical model. This approach opens a door to numerous applications for visualization, computation, and study of forestry and vegetation in urban as well as rural areas.
      PubDate: THU, 04 JAN 2024 09:17:23 -04
      Issue No: Vol. 17, No. null (2024)
       
  • SSNet: A Novel Transformer and CNN Hybrid Network for Remote Sensing
           Semantic Segmentation

    • Free pre-print version: Loading...

      Authors: Min Yao;Yaozu Zhang;Guofeng Liu;Dongdong Pang;
      Pages: 3023 - 3037
      Abstract: There are still various challenges in remote sensing semantic segmentation due to objects diversity and complexity. Transformer-based models have significant advantages in capturing global feature dependencies for segmentation. However, it unfortunately ignores local feature details. On the other hand, convolutional neural network (CNN), with a different interaction mechanism from transformer-based models, captures more small-scale local features instead of global features. In this article, a new semantic segmentation net framework named SSNet is proposed, which incorporates an encoder–decoder structure, optimizing the advantages of both local and global features. In addition, we build feature fuse module and feature inject module to largely fuse these two-style features. The former module captures the dependencies between different positions and channels to extract multiscale features, which promotes the segmentation precision on similar objects. The latter module condenses the global information in transformer and injects it into CNN to obtain a broad global field of view, in which the depthwise strip convolution improves the segmentation accuracy on tiny objects. A CNN-based decoder progressively recovers the feature map size, and a block called atrous spatial pyramid pooling is adopted in decoder to obtain a multiscale context. The skip connection is established between the decoder and the encoder, which retains important feature information of the shallow layer network and is conducive to achieving flow of multiscale features. To evaluate our model, we compare it with current state-of-the-art models on WHDLD and Potsdam datasets. The experimental results indicate that our proposed model achieves more precise semantic segmentation.
      PubDate: THU, 04 JAN 2024 09:17:23 -04
      Issue No: Vol. 17, No. null (2024)
       
  • S$^{2}$DCN: Spectral–Spatial Difference Convolution Network for
           Hyperspectral Image Classification

    • Free pre-print version: Loading...

      Authors: Zitong Zhang;Hanlin Feng;Chunlei Zhang;Qiaoyu Ma;Yuntao Li;
      Pages: 3053 - 3068
      Abstract: A novel spectral–spatial difference convolution network (S$^{2}$DCN) is proposed for hyperspectral image (HSI) classification, which integrates the difference principle into the deep learning framework. S$^{2}$DCN employs a learnable gradient encoding pattern to extract important detail features in spectral and spatial domains, alleviating the information loss caused by the oversmoothing effect in deep feature extraction. Specifically, the feature extraction modules in S$^{2}$DCN are designed, namely spectral difference convolution (SeDC) module and spatial difference convolution (SaDC) module. The SeDC module performs 1-D difference convolution in the spectral domain to capture peak-valley information in sensitive narrow bands, enhance subtle spectral differences, and preserve fine-grained features. The SaDC module employs 2-D difference convolution in the spatial domain, integrating fine-structural features while preserving the deep abstract features extracted by vanilla convolutions. This further empowers the capability of the model to extract discriminative features. A series of experiments are performed on four publicly available HSI datasets to demonstrate the effectiveness of S$^{2}$DCN method, which is compared with current state-of-the-art models. The experimental results show that the proposed S$^{2}$DCN outperforms competitors and achieves optimal classification performance.
      PubDate: WED, 03 JAN 2024 09:17:52 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Statistical Modeling of Polarimetric RCS of Road Surfaces for Scattering
           Simulation and Optimal Antenna Polarization Determination

    • Free pre-print version: Loading...

      Authors: Wietse Bouwmeester;Francesco Fioranelli;Alexander G. Yarovoy;
      Pages: 3078 - 3090
      Abstract: Incoherent backscattering of mm-waves from natural rough surfaces is considered. A novel method is proposed to determine the statistical properties of surface scattering from range profile measurements. The method is based on modeling the road surface as a grid of uncorrelated scattering elements, described by normalized scattering matrices. Using this model, expressions are derived to estimate the mean value and covariance matrix of surface scattering from measurement data. This procedure is then applied to measurement data of four road surface types, namely: 1) dry asphalt, 2) wet asphalt, 3) asphalt covered by basalt gravel, and 4) old asphalt. Using the derived statistical normalized radar cross-section models, two novel applications are proposed. First, a procedure for synthesizing/simulating surface clutter is proposed. This procedure is subsequently used to simulate received power from surfaces comprising patches of one or multiple road surface conditions. Excellent agreement between simulation and measurement results is demonstrated. Second, a method for determining the optimal polarization of the electromagnetic sensing waves used in a single-polarized radar system is proposed. This method is based on factorizing the antenna polarization vector into two bounded parameters, allowing for numerical evaluation of the minima and maxima for targets with a specified scattering matrix. This method is further extended to work with statistical descriptions of scattering matrices by means of Monte Carlo simulations.
      PubDate: THU, 04 JAN 2024 09:17:23 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Heterogeneous Dynamic Graph Convolutional Networks for Enhanced
           Spatiotemporal Flood Forecasting by Remote Sensing

    • Free pre-print version: Loading...

      Authors: Jiange Jiang;Chen Chen;Yang Zhou;Stefano Berretti;Lei Liu;Qingqi Pei;Jianming Zhou;Shaohua Wan;
      Pages: 3108 - 3122
      Abstract: Accurate and timely flood forecasting, facilitated by remote sensing technology, is crucial to mitigate the damage and loss of life caused by floods. However, despite years of research, accurate flood prediction still faces numerous challenges, including complex spatiotemporal features and varied flood patterns influenced by multiple variables. Moreover, long-term flood forecasting is always tricky due to the constantly changing conditions of the surrounding environment. In this study, we propose a heterogeneous dynamic temporal graph convolutional network (HD-TGCN) for flood forecasting. Specifically, we designed a dynamic temporal graph convolution module (D-TGCM) to generate a dynamic adjacency matrix by incorporating a multihead self-attention mechanism, enabling our model to capture the dynamic spatiotemporal features of flood data by utilizing temporal graph convolution operations on the dynamic matrix. Furthermore, to reflect the impact of multiple meteorological and hydrological features on the heterogeneity of flood data, we propose a novel approach that utilizes multiple parallel D-TGCMs for processing heterogeneous graph data and implements a fusion mechanism to capture varied flood patterns influenced by multiple variables. Experiments conducted on a real dataset in Wuyuan County, Jiangxi Province, demonstrate that the HD-TGCN outperforms the state-of-the-art flood prediction models in mean absolute error, Nash–Sutcliffe efficiency, and root-mean-square error, with improvements of 80.32%, 0.15%, and 73.99%, respectively, providing a more accurate flood forecasting method that will play a critical role in future flood disaster prevention and control.
      PubDate: WED, 03 JAN 2024 09:17:52 -04
      Issue No: Vol. 17, No. null (2024)
       
  • TCNet: Multiscale Fusion of Transformer and CNN for Semantic Segmentation
           of Remote Sensing Images

    • Free pre-print version: Loading...

      Authors: Xuyang Xiang;Wenping Gong;Shuailong Li;Jun Chen;Tianhe Ren;
      Pages: 3123 - 3136
      Abstract: Semantic segmentation of remote sensing images plays a critical role in areas such as urban change detection, environmental protection, and geohazard identification. Convolutional Neural Networks (CNNs) have been excessively employed for semantic segmentation over the past few years; however, a limitation of the CNN is that there exists a challenge in extracting the global context of remote sensing images, which is vital for semantic segmentation, due to the locality of the convolution operation. It is informed that the recently developed Transformer is equipped with powerful global modeling capabilities. A network called TCNet is proposed in this article, and a parallel-in-branch architecture of the Transformer and the CNN is adopted in the TCNet. As such, the TCNet takes advantage of both Transformer and CNN, and both global context and low-level spatial details could be captured in a much shallower manner. In addition, a novel fusion technique called Interactive Self-attention is advanced to fuse the multilevel features extracted from both branches. To bridge the semantic gap between regions, a skip connection module called Windowed Self-attention Gating is further developed and added to the progressive upsampling network. Experiments on three public datasets (i.e., Bijie Landslide Dataset, WHU Building Dataset, and Massachusetts Buildings Dataset) depict that TCNet yields superior performance over state-of-the-art models. The IoU values obtained by TCNet for these three datasets are 75.34% (ranked first among 10 models compared), 91.16% (ranked first among 13 models compared), and 76.21% (ranked first among 13 models compared), respectively.
      PubDate: THU, 04 JAN 2024 09:17:23 -04
      Issue No: Vol. 17, No. null (2024)
       
  • A New Extraction Method of Surface Water Based on Dense Time-Sequence
           Images

    • Free pre-print version: Loading...

      Authors: Hanyuan Liu;Yue Shi;Qinnan Chang;Rufat Guluzade;Xin Pan;Nan Xu;Penghua Hu;Xuechun Kong;Yingbao Yang;
      Pages: 3151 - 3166
      Abstract: Fluctuations in the surface water are indicators of climatic and biological environmental variations. The water index method is the predominant approach for water extraction owing to its simplicity of operation and high efficiency. Recognizing the limitations of individual water indices in extracting water over dense time-sequences, this study introduces a combined water index (CWI) frequency method to improve the water extraction results. The research findings indicate the following: 1) CWI demonstrates superior extraction accuracy for various types of water when compared with other water indices, underscoring its higher precision and broader applicability. 2) By integrating CWI with the water frequency method, we propose an effective approach for dynamically monitoring water. This method accurately reflects changes in water under different conditions within dense time-sequence images. 3) Our results highlight the method's ability to precisely monitor dynamic water changes, efficiently extract various water types from Sentinel-2 data, and its potential for large-scale surface water mapping applications.
      PubDate: MON, 01 JAN 2024 09:18:09 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Detail Enhanced Change Detection in VHR Images Using a Self-Supervised
           Multiscale Hybrid Network

    • Free pre-print version: Loading...

      Authors: Dalong Zheng;Zebin Wu;Jia Liu;Chih-Cheng Hung;Zhihui Wei;
      Pages: 3181 - 3196
      Abstract: The integration of the transformer and convolutional neural network (CNN) has become a useful method for change detection in remote sensing images. The main function of the transformer is to capture the global features, while the CNN is more for obtaining the local features. However, such an integration is not efficient for change detection in the very-high-resolution (VHR) remote sensing images with fine surface detail information. Hence, to improve this traditional construction of the transformer and CNN, we propose a dense Swin-Transformer-V2 (DST) and VGG16, coined as DST-VGG, for extracting the discriminatory features for change detection. The difference between our proposed network and other networks is that the output of the VGG16 encoders will be used in the DST in which more Swin-V2 blocks are added for fine feature extraction. The learning model in the VGG16 encoders employs a self-supervised method, which is guided through the change in details. Our network not only inherits the advantages of the integration of the transformer and CNN, but also captures the features of change relationship through the DST and catches the primitive features in both prechanged and postchanged regions through the VGG16. In addition, we design a mixed feature pyramid within the DST, which provides interlayer interaction information and intralayer multiscale information for a more complete feature learning within the new network. Furthermore, we impose a self-supervised strategy to guide the VGG16 provide the semantic change information from the output features of the encoder. We compared our experimental results with those of the state-of-the-art methods on four commonly used public VHR remote sensing datasets. It shows that our network performs better, in terms of F1, IoU, and OA, than those of the existing networks for change detection.
      PubDate: MON, 01 JAN 2024 09:18:11 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Unsupervised Domain Adaptation With Debiased Contrastive Learning and
           Support-Set Guided Pseudolabeling for Remote Sensing Images

    • Free pre-print version: Loading...

      Authors: Debojyoti Biswas;Jelena Tešić;
      Pages: 3197 - 3210
      Abstract: The variability in different altitudes, geographical variances, and weather conditions across datasets degrade state-of-the-art (SOTA) deep neural network object detection performance. Unsupervised and semisupervised domain adaptations (DAs) are decent solutions to bridge the gap between two different distributions of datasets. The SOTA pseudolabeling process is susceptible to background noise, hindering the optimal performance in target datasets. The existing contrastive DA methods overlook the bias effect introduced from the false negative (FN) target samples, which mislead the complete learning process. This article proposes support-guided debiased contrastive learning for DA to properly label the unlabeled target dataset and remove the bias toward target detection. We introduce: 1) a support-set curated approach to generate high-quality pseudolabels from the target dataset proposals; 2) a reduced distribution gap across different datasets using domain alignment on local, global, and instance-aware features for remote sensing datasets; and 3) novel debiased contrastive loss function that makes the model more robust for the variable appearance of a particular class over images and domains. The proposed debiased contrastive learning pivots on class probabilities to address the challenge of FNs in the unsupervised framework. Our model outperforms the compared SOTA models with a minimum gain of +3.9%, +3.2%, +12.7%, and +2.1% of mean average precision for DIOR, DOTA, Visdrone, and UAVDT datasets, respectively.
      PubDate: THU, 04 JAN 2024 09:17:23 -04
      Issue No: Vol. 17, No. null (2024)
       
  • BCNet: Background Conversion Network for SAR Data Generation

    • Free pre-print version: Loading...

      Authors: Jiawei Luan;Zhong Xu;Bowen Li;Jinshan Ding;
      Pages: 3211 - 3225
      Abstract: Collecting large-scene synthetic aperture radar (SAR) images with targets of interest (TOI) has been a challenging task. To embed TOI slices into measured large scenes can be a good solution. Current methods for SAR TOI slice generation are mainly based on a single data source. Poor background variability of generated images leads to difficulty in naturally embedding TOI slices into large scenes. This article presents a SAR target background conversion network (BCNet), which combines TOI slices with large-scene slices under the same operating condition. The background of TOI slices is converted to large-scene background while preserving the target scattering characteristics to enhance the target background diversity and variability. Background conversion is a special image style transfer, and BCNet uses CycleGAN as the baseline model. The problem that the baseline model may result in target missing is analyzed by Bayesian theory, and then, a new loss function Bysloss is designed to preserve the characteristics of target shadow and scattering center. A new image fusion module has been developed to generate training data for robust background conversion. In addition, the generated high-quality background conversion images are used for two-way recognition performance verification, large scene, and generated TOI slices fusion verification, respectively. The experimental results have shown that the generated data can be successfully used for SAR automatic target recognition in few-shot conditions, and also have strong potential in generating large-scene SAR images with TOI, SAR deception jamming, and augmenting the target detection dataset.
      PubDate: TUE, 09 JAN 2024 09:18:02 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Correction of the SMAP Sea Surface Brightness Temperature and Retrieval of
           Sea Surface Salinity Incorporating CYGNSS Observables

    • Free pre-print version: Loading...

      Authors: Zheng Li;Fei Guo;Zhiyu Zhang;Xiaohong Zhang;
      Pages: 3226 - 3235
      Abstract: The correction of sea surface brightness temperature is crucial for improving the accuracy of sea surface salinity (SSS) retrieval by L-band microwave radiometer. However, the traditional method of correcting brightness temperature using only wind speed and significant wave height (SWH) is inadequate, as sea surface roughness is affected by multiple factors. The Global Navigation Satellite System Reflectometer (GNSS-R) observables, which directly respond to sea surface roughness, have been preliminarily validated in ground-based experiments for their potential to correct sea surface brightness temperature. Compared with ground-based GNSS-R, spaceborne GNSS-R has a wider coverage and can better support the brightness temperature correction of spaceborne L-band microwave radiometers. This article has preliminarily verified the correlation between cyclone GNSS (CYGNSS) observables and brightness temperature variations, and found that the incidence angle of the observable needs to be taken into account when retrieving SSS jointly with soil moisture active and passive (SMAP) and CYGNSS. A multilayer perceptron (MLP) model was established to assess the SSS retrieval performance of SMAP combined with different parameters. The results show that the retrieval performance based on the MLP model is better than that based on the geophysical model function model. Compared with joint wind speed and SWH, joint CYGNSS observables performs better in retrieving SSS. The root mean square error of retrieval salinity decreased from 0.58 to 0.46 psu, and the correlation coefficient (R) increased from 0.83 to 0.90. This provides reference for future joint retrieval of SSS using L-band microwave radiometers and spaceborne GNSS-R.
      PubDate: MON, 01 JAN 2024 09:18:10 -04
      Issue No: Vol. 17, No. null (2024)
       
  • An Unsupervised Transformer-Based Multivariate Alteration Detection
           Approach for Change Detection in VHR Remote Sensing Images

    • Free pre-print version: Loading...

      Authors: Yizhang Lin;Sicong Liu;Yongjie Zheng;Xiaohua Tong;Huan Xie;Hongming Zhu;Kecheng Du;Hui Zhao;Jie Zhang;
      Pages: 3251 - 3261
      Abstract: Multitemporal change detection (CD) plays a crucial role in the remote sensing application field. In recent years, supervised deep learning methods have shown excellent performance in detecting changes in very-high-resolution (VHR) images. However, these methods require a large number of labeled samples for training, making the process time-consuming and labor-intensive. Unsupervised approaches are more attractive in practical applications since they can produce a CD map without relying on any ground reference or prior knowledge. In this article, we propose a novel unsupervised CD approach, named transformer-based multivariate alteration detection (trans-MAD). It utilizes a pre-detection strategy that combines the compressed change vector analysis and the iteratively reweighted multivariate alteration detection (IR-MAD) to generate reliable pseudotraining samples. More accurate and robust CD results can be achieved by leveraging the IR-MAD to detect insignificant changes and by incorporating the transformer-based attention mechanism to model the difference or similarity between two distant pixels in an image. The proposed trans-MAD approach was validated on two VHR bitemporal satellite remote sensing datasets, and the obtained experimental results demonstrated its superiority comparing with the state-of-the-art unsupervised CD methods.
      PubDate: THU, 04 JAN 2024 09:17:23 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Kernel Approximation on a Quantum Annealer for Remote Sensing Regression
           Tasks

    • Free pre-print version: Loading...

      Authors: Edoardo Pasetto;Morris Riedel;Kristel Michielsen;Gabriele Cavallaro;
      Pages: 3262 - 3269
      Abstract: The increased development of quantum computing hardware in recent years has led to increased interest in its application to various areas. Finding effective ways to apply this technology to real-world use-cases is a current area of research in the remote sensing community. This article proposes an adiabatic quantum kitchen sinks (AQKS) kernel approximation algorithm with parallel quantum annealing on the D-Wave Advantage quantum annealer. The proposed implementation is applied to support vector regression and Gaussian process regression algorithms. To evaluate its performance, a regression problem related to estimating chlorophyll concentration in water is considered. The proposed algorithm was tested on two real-world datasets and its results were compared with those obtained by a classical implementation of kernel-based algorithms and a random kitchen sinks implementation. On average, the parallel AQKS achieved comparable results to the benchmark methods, indicating its potential for future applications.
      PubDate: FRI, 05 JAN 2024 09:18:18 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Progressive Feature Fusion Framework Based on Graph Convolutional Network
           for Remote Sensing Scene Classification

    • Free pre-print version: Loading...

      Authors: Chongyang Zhang;Bin Wang;
      Pages: 3270 - 3284
      Abstract: Remote sensing (RS) scene classification plays an important role in the intelligent interpretation of RS data. Recently, convolutional neural network (CNN)-based and attention-based methods have become the mainstream of RS scene classification with impressive results. However, existing CNN-based methods do not utilize long-range information, and existing attention-based methods do not fully exploit multiscale information, although both aspects of information are essential for a comprehensive understanding of RS scene images. To overcome the above limitations, we propose a progressive feature fusion (PFF) framework based on graph convolutional network (GCN), namely PFFGCN for RS scene classification in this article, which has a strong ability to learn both multiscale and contextual (local/long-range) information in RS scene images. It mainly consists of two modules: a multilayer feature extraction module and a multiscale contextual information fusion (MCIF) module. The MFE module is utilized to extract multilevel features and global features, and the MCIF module is constructed to capture rich contextual information from multilevel features and fuse them in a progressive manner. In MCIF, GCN is adopted to explore intrinsic attributes (including the topological structure and the contextual information) hidden in each feature map. Through the PFF strategy, the graph features at each level are fused with the next-level features to reduce the semantic gap between nonadjacent features and enhance the multiscale representation of the model. Besides, grouped GCN based on channel grouping is further proposed to improve the efficiency of PFFGCN. The proposed method is extensively evaluated on various RS scene classification datasets, and the experimental results demonstrate that the proposed method outperforms current state-of-the-art methods.
      PubDate: FRI, 05 JAN 2024 09:18:18 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Forecast of Ionospheric TEC Maps Using ConvGRU Deep Learning Over China

    • Free pre-print version: Loading...

      Authors: Jun Tang;Zhengyu Zhong;Mingfei Ding;Dengpan Yang;Heng Liu;
      Pages: 3334 - 3344
      Abstract: In this article, we propose a convolutional gated recurrent unit (ConvGRU) deep learning method to forecast ionospheric total electron content (TEC) over China based on the regional ionospheric maps (RIMs) from 2015 to 2018. First, we use Global Navigation Satellite System observations from the Crustal Movement Observation Network of China to generate the RIMs of China (CRIMs). Second, we use the CRIMs of 2015–2017 as the training set to predict the ionospheric TEC over China in 2018. Finally, comparative experiments are carried out with ConvLSTM, International Reference Ionosphere (IRI), and Center for Orbit Determination in Europe's (CODE's) 1-day predicted Global Ionospheric Map (C1PG) released by CODE. In addition, we add geomagnetic indices (ap, Kp, and Dst) and solar activity index (F10.7) as the training set to analyze the prediction accuracy of the model (using -A if there are no indices, and -B if there are indices). The results illustrate that the prediction accuracy of ConvLSTM-B and ConvGRU-B models are improved on both geomagnetic storm and quiet days, and the improvement is more obvious on geomagnetic storm days. Furthermore, the root mean square error (RMSE) of the ConvGRU-B model decreases by 28%, 22.4%, and 5.9% compared to that of the ConvGRU-A, IRI-2016, and ConvLSTM-B models during geomagnetic storm days, respectively. For the prediction accuracy of a certain grid point, the RMSE of the ConvGRU-B model decreases by 23%, 32.6%, and 19.3% during geomagnetic quiet days and 24.4%, 30.6%, and 15.7% during geomagnetic storm days compared to that of the ConvGRU-A, IRI-2016, and ConvLSTM-B models, respectively. For the forecast accuracy of TEC in different seasons, the performance of the ConvGRU-B model is also better than that of the ConvLSTM-B model in 2018. These results show that the ConvGRU-B model has competitive performance in RIMs prediction over China during the geomagnetic quiet and storm days.
      PubDate: WED, 03 JAN 2024 09:17:52 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Mask-Guided Local–Global Attentive Network for Change Detection in
           Remote Sensing Images

    • Free pre-print version: Loading...

      Authors: Fengchao Xiong;Tianhan Li;Jingzhou Chen;Jun Zhou;Yuntao Qian;
      Pages: 3366 - 3378
      Abstract: Change detection in remote sensing images is a challenging task due to object appearance diversity and the interference of complex backgrounds. Self-attention- and spatial-attention-based solutions face limitations, such as high memory consumption and an inadequate ability to capture long-range relations, leading to imprecise contextual information and restricted performance. To address these challenges, this article introduces a novel mask-guided local–global attentive network (MLA-Net). The MLA-Net incorporates a memory-efficient local–global attention module that leverages the benefits of both self-attention and spatial attention to accurately capture the local–global context. Through simultaneous exploitation of context within inter- and intrapatches and information refinement, the feature representation capability is significantly enhanced. In addition, we introduce a change mask to refine feature differences and eliminate interference from irrelevant changes caused by complex backgrounds. Accordingly, a mask loss is defined to guide the generation of the mask. Extensive experiments on the LEVIR-CD, WHU-CD, and CLCD datasets show that our MLA-Net performs better than state-of-the-art methods.
      PubDate: FRI, 05 JAN 2024 09:18:18 -04
      Issue No: Vol. 17, No. null (2024)
       
  • A Restoration Scheme for Spatial and Spectral Resolution of the
           Panchromatic Image Using the Convolutional Neural Network

    • Free pre-print version: Loading...

      Authors: Xin Jin;Ling Liu;Xiaoxuan Ren;Qian Jiang;Shin-Jye Lee;Jun Zhang;Shaowen Yao;
      Pages: 3379 - 3393
      Abstract: Remote sensing images are the product of information obtained by various sensors, and the higher the resolution of the image, the more information it contains. Therefore, improving the resolution of the remote sensing image is conducive to identify Earth resources from the remote sensing image. In this article, we present a multiple-branch panchromatic image resolution restoration network based on the convolutional neural network to improve the spatial and spectral resolution of the panchromatic image simultaneously, named MBPRR-Net. Specifically, we adopt a multibranch structure to extract abundant features and utilize a feature channel mixing block to enhance the interaction of adjacent channels between features. Feature aggregation in our method is used to learn more effective features from each branch, and then a cubic filter is utilized to enhance the aggregated features. After feature extraction, we use a recovery architecture to generate the final image. Moreover, we utilize image super-resolution to restore spatial resolution and image colorization to restore the spectral resolution so that we can compare it with some image colorization and super-resolution methods to verify the proposed method. Experiments show that the performance of our method is outstanding in terms of visual effects and objective evaluation metrics compared with some existing excellent image super-resolution and colorization methods.
      PubDate: TUE, 09 JAN 2024 09:18:02 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Improved Few-Shot SAR Image Generation by Enhancing Diversity

    • Free pre-print version: Loading...

      Authors: Jianghan Bao;Wen Ming Yu;Kaiqiao Yang;Che Liu;Tie Jun Cui;
      Pages: 3394 - 3408
      Abstract: Due to their remarkable capabilities of generation, deep-learning-based (DL) generative models have been widely applied in the field of synthetic aperture radar (SAR) image synthesis. This kind of data-driven DL methods usually requires abundant training samples to guarantee the performance. However, the number of SAR images for training is often insufficient because of expensive acquisitions. This typical few-shot image generation (FSIG) task still remains not fully investigated. In this article, we propose an optical-to-SAR (O2S) image translation model with a pairwise distance (PD) loss to enhance the diversity of generation. First, we replace the semantic maps used as the input of network in previous studies with more easily available optical images and apply the popular pix2pix model in image-to-image translation tasks as the foundation network. Second, inspired by the FSIG works in the traditional computer vision field, we propose a similarity preservation term in the loss function, which encourages the generated images to inherit the similarity relationship of the corresponding simulated SAR images. Third, the data augmentation experiments on the MSTAR dataset indicates the effectiveness of our model. With only five samples for each target category and six categories in total, the basic O2S network boosts the classification accuracy by 4.81% and 2.27% for data of depression angle of 15° and 17°, respectively. The PD loss is capable of bringing additional 2.23% and 1.78% improvement. The investigation on similarity curves also suggests that the generated images enhanced by the PD loss have closer similarity behaviors to the real SAR images.
      PubDate: WED, 10 JAN 2024 09:17:23 -04
      Issue No: Vol. 17, No. null (2024)
       
  • A Velocity Ambiguity Resolution Algorithm Based on Improved Hypothetical
           Phase Compensation for TDM-MIMO Radar Traffic Target Imaging

    • Free pre-print version: Loading...

      Authors: Bo Yang;Siqi Liu;Hua Zhang;Yongjun Zhou;
      Pages: 3409 - 3424
      Abstract: In principle, the imaging millimeter-wave radar based on time-division multiplexing multiple-input multiple-output (MIMO) technologies can provide richer target information for intelligent transportation systems due to the high-density target point cloud output. However, the chirp repetition interval extension reduces radar's inherent maximum detectable velocity, leading to the unavoidable problem of estimating the target velocity with a large ambiguity period in imaging radar moving target surveillance applications. To alleviate these problems, we propose an improved hypothetical phase compensation algorithm. Unlike the original method of determining the Doppler ambiguity period by comparing the peak amplitude of the angular power spectrum in each hypothetical case, the proposed algorithm selects the peak of the angular power signal-to-noise ratio (SNR) spectrum as the processing object and jointly decides the target speed by the average of the two highest wave peak intervals in the SNR variation curve. Simulations and practical experiments show that the improved algorithm has higher anti-interference performance. In particular, the proposed algorithm can remain continuously effective when multiple targets or angle information exist in the same distance Doppler cell, making it more suitable for MIMO imaging applications.
      PubDate: WED, 10 JAN 2024 09:17:21 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Monthly NDVI Prediction Using Spatial Autocorrelation and Nonlocal
           Attention Networks

    • Free pre-print version: Loading...

      Authors: Lei Xu;Ruinan Cai;Hongchu Yu;Wenying Du;Zeqiang Chen;Nengcheng Chen;
      Pages: 3425 - 3437
      Abstract: Accurate prediction of vegetation indices is useful for helping maintain vegetation stability, sustaining food production, and reducing socioeconomic losses. The traditional convolutional long short-term memory (ConvLSTM) model for vegetation prediction ignores the spatial aggregation characteristics of the normalized difference vegetation index (NDVI) itself and the global dependence information in space. In this study, we propose a new NDVI prediction method, namely, the ConvLSTM with spatial autocorrelation and nonlocal attention module (ConvLSTM-SAC-NL), by combining the nonlocal attention module to capture long-range dependence and the spatial autocorrelation modeling based on the local Moran index to learn spatial dependence. The experimental results indicate that the ConvLSTM-SAC-NL model outperforms seven baseline forecasting models, with an R${}^{2}$ of 0.881 in monthly NDVI prediction in the Huangpi district of Wuhan city, relative to the R${}^{2}$ values of 0.758, 0.777, 0.741, 0.776, 0.804, 0.829, and 0.815 for random forest, support vector machine regression, long short-term memory, bidirectional long short-term memory, graph convolutional network, predictive recurrent neural network, and ConvLSTM models, respectively. Spatially, the prediction results of the ConvLSTM-SAC-NL model demonstrate improved accuracy over 91.49$\%$ of the study area when compared with ConvLTSM. Therefore, the proposed ConvLSTM-SAC-NL model could serve as an effective approach for short-term prediction of vegetation conditions at regional scales.
      PubDate: FRI, 05 JAN 2024 09:18:18 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Agreement Analysis and Accuracy Assessment of Multiple Mangrove Datasets
           in Guangxi Beibu Gulf and Guangdong-Hong Kong-Macau Greater Bay, China,
           for 2000–2020

    • Free pre-print version: Loading...

      Authors: Zhijie Xiao;Weiguo Jiang;Zhifeng Wu;Ziyan Ling;Yawen Deng;Ze Zhang;Kaifeng Peng;
      Pages: 3438 - 3451
      Abstract: Accurate and reliable mangrove datasets are essential for the protection and management of mangrove ecosystems. Therefore, the evaluation of the current mangrove datasets and understanding the differences among them are critical. This study takes the Guangxi Beibu Gulf (GBG) and Guangdong-Hong Kong-Macao Greater Bay Area (GBA) as the study areas and analyzes the agreement and accuracy of eight mangrove datasets from 2000 to 2020 using area comparison, spatial agreement analysis, and absolute accuracy evaluation. The results show that; 1) significant differences exist in mangrove area and spatial distribution among the different mangrove datasets, with the percentage of high agreement areas ranging from 10% to 42%. 2) The overall accuracy of the evaluated mangrove datasets ranges from 56.2% to 95.6%, and the classification accuracy of mangrove datasets in inland areas is lower than the overall level. 3) There are regional differences in the quality of mangrove datasets, with the agreement and accuracy of mangrove datasets in the GBG being greater than those in the GBA. 4) Fine-scale mangrove mapping based on high-resolution remote sensing images, such as Sentinel, and global mangrove mapping based on the Google Earth Engine (GEE) cloud platform should be emphasized in the future. The findings of this study can provide guidance for data users to select appropriate mangrove datasets and a reference for future mangrove mapping research.
      PubDate: FRI, 12 JAN 2024 09:16:33 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Comparison and Evaluation of TV-Based and Low-Rank-Based Destriping
           Algorithms for Hyperspectral Imagery

    • Free pre-print version: Loading...

      Authors: Fan Shen;Jianyu Chen;Yuntao Cheng;Qiankun Zhu;
      Pages: 3452 - 3470
      Abstract: Remote sensing image data especially hyperspectral imagery (HSI) data have been widely applied in various research fields. However, HSI often suffers from significant noise contamination, such as stripe noise and Gaussian noise. These noises adversely affect the utilization of remote sensing data, prompting the development of numerous noise removal algorithms for HSI. Specifically, addressing stripe noise involves employing different strategies, including the employment of total-variation-based (TV-based) and low-rank-based algorithms. Yet, evaluating the effectiveness and applicability of these algorithms can be challenging due to variations in testing conditions provided by their respective authors during the proposal phase. Consequently, our aim was to offer a comprehensive and impartial evaluation of these stripe noise removal algorithms for HSI. Within the TV-based and low-rank-based algorithm categories, we had chosen 10 distinct algorithms on which third-party testing and evaluation would be conducted. We considered various noise scenarios, resulting in a total of 48 noise configurations. Our evaluation encompassed multiple dimensions, including the quality of noise removal, computational speed, and parameter stability, thereby delivering a comprehensive assessment of the performance of different stripe noise removal algorithms. In addition, we conducted an analysis to identify the underlying causes of varying denoising results across different algorithms. This analysis offered valuable insights and recommendations for the future development of denoising algorithms.
      PubDate: WED, 10 JAN 2024 09:17:23 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Open Set Recognition and Category Discovery Framework for SAR Target
           Classification Based on K-Contrast Loss and Deep Clustering

    • Free pre-print version: Loading...

      Authors: Mingyao Chen;Jing-Yuan Xia;Tianpeng Liu;Li Liu;Yongxiang Liu;
      Pages: 3489 - 3501
      Abstract: Synthetic aperture radar automatic target recognition (SAR ATR) has been widely studied in recent years. Most ATR models are designed based on the traditional closed-set assumption. This type of ATR model can only identify target categories existing in the training set, and it will result in missed detection or misclassification of unseen target categories encountered in battlefield reconnaissance, posing a potential threat. Therefore, it is of great significance to design a model that can simultaneously achieve known class classification and unknown class judgment. In addition, researchers usually use the obtained unknown class data for model relearning to enable it to recognize new categories. However, before this process, it is necessary to manually interpret and annotate the obtained unknown class data, which undoubtedly requires a large time cost and is difficult to meet the timeliness requirements. To solve these problems, we propose a framework that integrates the open-set recognition module and the novel class discovery module. By introducing the K-contrast loss, the open-set recognition module can accurately distinguish unknown class data, classify known class data, and then transfer the known class knowledge through deep clustering for clustering annotation of unknown class data. Extensive experimental results on the MSTAR benchmark dataset demonstrate the effectiveness of the proposed methods.
      PubDate: FRI, 12 JAN 2024 09:16:33 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Ellipse Polar Encoding for Oriented SAR Ship Detection

    • Free pre-print version: Loading...

      Authors: Jun Liu;Lin Liu;Jiarong Xiao;
      Pages: 3502 - 3515
      Abstract: Ship detection in synthetic aperture radar (SAR) images using deep neural networks often relies on horizontal bounding box, which fail to capture ship orientation and aspect ratio accurately. To address this limitation, oriented ship detection methods based on oriented bounding box (OBB) are gaining attention. However, most available existing OBB methods suffer from boundary discontinuity problem, leading to convergence problems and unsatisfied orientation detection performance. In this article, we propose an innovative oriented SAR ship detection method using ellipse polar encoding (EPE). By representing the ship detection box as an ellipse and employing a set of vectors from the center to the boundary as encoded parameters, our method exhibits smooth variations, enhances convergence, and efficiently decodes into OBB. We further develop a lightweight and effective oriented SAR ship detection network based on this methodology. To account for SAR image characteristics, such as speckle and deformation causing deviations from true ellipses, we introduce an intersection over union weighted EPE loss. The experimental results on the rotated ship detection dataset in SAR images demonstrate the effectiveness of our proposed method in significantly improving detection performance compared with other oriented target detection methods.
      PubDate: WED, 10 JAN 2024 09:17:21 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Poverty Estimation Using a ConvLSTM-Based Model With Multisource Remote
           Sensing Data: A Case Study in Nigeria

    • Free pre-print version: Loading...

      Authors: Jie Tang;Xizhi Zhao;Fuhao Zhang;Agen Qiu;Kunwang Tao;
      Pages: 3516 - 3529
      Abstract: Poverty is a global challenge, the effects of which are felt on the individual to national scale. To develop effective support policies to reduce poverty, local governments require precise poverty distribution data, which are lacking in many areas. In this study, we proposed a model to estimate poverty on a spatial scale of 10 × 10 km by combining features extracted from multiple data sources, including nighttime light remote sensing data, normalized difference vegetation index, surface reflectance, land cover type, and slope data, and applied the model to Nigeria. Considering that the trends of environmental factors contain valid information related to poverty, time-series features were extracted through convolutional long short-term memory and used for the assessment. The poverty level is represented by the wealth index derived from the Demographic and Health Survey Program. The model exhibited good ability to estimate poverty, with an R2 of 0.73 between the actual and estimated wealth index in Nigeria in 2018. Applying the proposed model to poverty estimation for Nigeria in 2021 yielded an R2 value of 0.69, indicating good generalization ability. To further validate model reliability, we compared the assessment results with high-resolution satellite imagery and a state-level multidimensional poverty index. We also investigated the impact of incorporating time-series features on the accuracy of poverty assessment. Results showed that the addition of time-series features increased the accuracy of poverty estimation from 0.64 to 0.73. The proposed method has valuable applications for estimating poverty at the grid scale in countries without such data.
      PubDate: MON, 15 JAN 2024 09:20:19 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Lightweight Attention-Guided YOLO With Level Set Layer for Landslide
           Detection From Optical Satellite Images

    • Free pre-print version: Loading...

      Authors: Yueheng Yang;Zelang Miao;Hua Zhang;Bing Wang;Lixin Wu;
      Pages: 3543 - 3559
      Abstract: Landslide inventory is significant for landslide disaster reduction. To construct the landslide inventory, deep learning has received growing attention to detect landslides from satellite images. Among various deep learning algorithms, you-only-look-once (YOLO) has a strong ability to detect objects efficiently and has been widely used in landslide extraction. Despite its efficiency, there is no general rule to select the backbone and attention mechanism for YOLO. The selection of these two modules depends on specific application needs. Meanwhile, YOLO output is a series of anchor boxes, not accurate landslide boundaries. A single bounding box may contain many landslides and cannot extract individual landslides, limiting the YOLO applications in constructing landslide inventory. To address these issues, this article presents a lightweight attention-guided YOLO with level set layer (LA-YOLO-LLL) for landslide detection from optical satellite images. First, we introduced the MobileNetv3 to replace the original backbone of YOLO to simultaneously reduce the parameter complexity and improve the model transferability. Then, we presented a light pyramid features reuse fusion attention mechanism to improve landslide detection performance. Finally, we integrated the level set layer into YOLO head to produce accurate landslide boundaries. This article validated the accuracy and transferability of the presented method in two study areas (Bijie and Taiwan) with similar geo-environmental conditions. Experimental results show that the presented LA-YOLO-LLL model outperformed traditional YOLO in landslide detection. Findings in this article are valuable for landslide inventory construction, land use planning and risk control.
      PubDate: TUE, 09 JAN 2024 09:18:09 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Multilevel Pyramid Feature Extraction and Task Decoupling Network for SAR
           Ship Detection

    • Free pre-print version: Loading...

      Authors: Yanshan Li;Wenjun Liu;Ruo Qi;
      Pages: 3560 - 3570
      Abstract: Synthetic aperture radar (SAR) target detection plays a crucial role in both military and civilian fields, attracting significant attention from researchers globally. CenterNet, a single-stage target detection method, is known for its high detection speed and accuracy by eliminating anchor-related calculations and nonmaximum suppression. However, directly applying CenterNet to SAR ship detection poses challenges due to the distinctive characteristics of SAR images, including lower resolution, lower signal-to-noise ratio, and larger ship aspect ratios. To address these challenges, we propose MPDNet. which introduces a multilevel pyramid feature extraction module (MP-FEM) to replace the encoding–decoding structure in CenterNet. MP-FEM employs multilevel pyramid and channel compression to fuse multiscale SAR image features and acquire deep features quickly. Second, we propose the convolution channel attention module, which improves the multilayer perceptron in the common pooling attention mechanism into a multistage and 1-D convolution. Therefore, the feature extraction capability of MP-FEM is further refined. Furthermore, we propose the detection task decoupling module (DTDM), which considers the characteristics of SAR ships and effectively detects smaller targets of different sizes, distinguishing the centers and sizes of densely arranged ships. DTDM extracts task-related features from the original feature map before inputting it into the three detection headers, thereby addressing the problem of task coupling in CenterNet's detection header module for SAR ship detection. Finally, the experimental results on SSDD dataset and SAR-ship-dataset show that the proposed network can significantly improve the SAR target detection accuracy.
      PubDate: WED, 17 JAN 2024 09:16:24 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Comparison of Scanning Strategies in UAV-Mounted Multichannel GPR-SAR
           Systems Using Antenna Arrays

    • Free pre-print version: Loading...

      Authors: María García-Fernández;Guillermo Álvarez-Narciandi;Fernando Las Heras;Yuri Álvarez-López;
      Pages: 3571 - 3586
      Abstract: Ground penetrating radar (GPR) systems on board unmanned aerial vehicles (UAVs) have been successfully used for subsurface imaging applications. Their capability to detect buried targets avoiding the contact with the soil turn these systems into a great solution to detect buried threats, such as landmines and improvised explosive devices. Significant advances have been also conducted to enhance the detection capabilities of these systems, complementing the synthetic aperture radar (SAR) processing methods with several clutter mitigation techniques. However, the improvement in the scanning throughput (i.e., increasing the inspected area in a given time) remains a significant challenge. In this regard, this article compares several scanning strategies for UAV-mounted multichannel GPR-SAR systems using antenna arrays. In particular, two different scanning strategies have been compared: a uniform scheme and a nonuniform strategy called 3X. In addition, different across-track spacing values to generate dense and sparse sampling distributions were considered for each scanning scheme. After conducting a theoretical analysis of these strategies, they have been experimentally validated with measurements gathered with a portable scanner and during flights in realistic scenarios. Results show that the dense configurations of both scanning strategies yield good quality images of buried targets while improving the scanning throughput (compared to a single-channel architecture). In particular, the dense uniform scheme (with a 20-cm across-track spacing) achieves a greater reduction in the inspection time, compared to the dense 3X strategy, at the expense of a slightly smaller signal to clutter ratio.
      PubDate: TUE, 09 JAN 2024 09:18:02 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Satellite Radar Altimetry Supporting Coastal Hydrology: Case Studies of
           Guadalquivir River Estuary and Ebro River Delta (Spain)

    • Free pre-print version: Loading...

      Authors: Jesús Gómez-Enri;A. Aldarias;R. Mulero-Martínez;S. Vignudelli;M. Bruno;R. Mañanes;A. Izquierdo;M. Fernández-Barba;
      Pages: 3587 - 3599
      Abstract: Coastal zones close to big river systems are hydrodynamically characterized by inland and open ocean waters. Satellite altimetry data have recently been scientifically exploited in coastal zones. However, more efforts are still needed in terms of using these data in new applications in the coastal strip. Our study analyzes the capability of the European Space Agency CryoSat-2 satellite to measure the sea level elevation associated with bulge-like lens of less salty waters from river discharges, spreading into the adjacent coastal zones. We analyze four events of high river freshwater discharges in two river estuaries in Spain: Guadalquivir and Ebro. We obtain along-track Absolute Dynamic Topography from CryoSat-2 satellite during these events and compare them with periods of low discharge conditions. In three of the events, the bulges increase the sea level between 5 and 10 cm with respect to the sea level observed in low / normal discharge conditions. The extension of the lens is far from the estuary mouth: more than 25 km from the coast in the case of the Guadalquivir River and about 50 km for the Ebro River. The wind regime and the surface circulation explain the lack of a higher sea level in one of the events analyzed.
      PubDate: THU, 04 JAN 2024 09:17:23 -04
      Issue No: Vol. 17, No. null (2024)
       
  • StHCFormer: A Multivariate Ocean Weather Predicting Method Based on
           Spatiotemporal Hybrid Convolutional Attention Networks

    • Free pre-print version: Loading...

      Authors: Lianlei Lin;Zongwei Zhang;Hangyi Yu;Junkai Wang;Sheng Gao;Hanqing Zhao;Jiaqi Zhang;
      Pages: 3600 - 3614
      Abstract: Ocean weather prediction is crucial for various applications, such as global climate prediction, marine environmental protection, and offshore production. However, current data-based marine weather prediction methods have limitations when predicting multiple variables in a particular area, failing to meet the efficiency and accuracy requirements of practical applications. In the realm of ocean weather variations, the presence of highly interconnected spatial and temporal continuations, coupled with the mutual influence of individual variables, underscores the utmost importance of effectively capturing dynamic correlations encompassing space, time, and variables to accurately predict ocean weather. To address this, we developed a novel approach called StHCFormer, which is a multivariate spatiotemporal hybrid convolutional attention network. The first key component of StHCFormer is the spatiotemporal hybrid convolutional attention (StHCA) module, which leverages a hybrid convolutional attention mechanism to explore both global spatial representations and local features. Additionally, the module incorporates temporal attention to capture the temporal dependence of weather records and effectively captures the dynamic correlations among multiple variables through channel deflation and weighted residuals. To ensure balanced variable losses, we introduced the concept of homoscedasticity uncertainty loss to dynamically adjust the multitask weights. This guarantees a global optimal solution and leads to more accurate multivariate ocean weather prediction. Finally, we conducted a comprehensive evaluation and comparison of the StHCFormer model with other state-of-the-art algorithms using the ERA5 dataset in the Philippine Sea. The results demonstrated that StHCFormer outperforms existing methods in marine multivariate field weather prediction.
      PubDate: TUE, 16 JAN 2024 09:16:10 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Analysis of C-Band Radar Temporal Coherence Over an Irrigated Olive
           Orchard in a Semi-Arid Region

    • Free pre-print version: Loading...

      Authors: Adnane Chakir;Pierre-Louis Frison;Ludovic Villard;Nadia Ouaadi;Pascal Fanise;Khabba Saïd;Valérie Le Dantec;Rafi Zoubair;Jamal Ezzahar;Bénédicte Fruneau;Jean-Paul Rudant;Jarlan Lionel;
      Pages: 3635 - 3647
      Abstract: This article aimed to monitor vegetation using C-band radar data at a subdaily time step. To this end, radar measurements using tower-mounted antennas with a 15-min time step, along with physiology-related information (sapflow and micrometric dendrometry), were acquired quasi-continuously from March 2020 to December 2021 in an olive orchard located near Marrakech, Morocco. The article focused on temporal coherence, whose clear diurnal cycle (highest at night and lowest at the end of the afternoon) had been highlighted over tropical and boreal forests in previous studies. The results showed that coherence was highly sensitive to: wind-induced movement of scatterers, since coherence was lowest when wind speed was highest in late afternoon, and vegetation activity, especially its water dynamics, since the morning coherence drop coincided with the onset of sapflow and the daily evapotranspiration cycle, as well as the good agreement between the temporal drop rate of coherence and the daily residual variation in trunk circumference (i.e., deviation from long-term trend). Finally, coherence remained high for temporal baselines of several days, showing that sentinel-1 data (when both satellites are operational) may be well suited for such studies, especially with acquisitions made during morning passes, when wind speed is low. These results open perspectives for monitoring tree crop physiology using high-revisit-time radar observations.
      PubDate: THU, 11 JAN 2024 09:15:59 -04
      Issue No: Vol. 17, No. null (2024)
       
  • BisDeNet: A New Lightweight Deep Learning-Based Framework for Efficient
           Landslide Detection

    • Free pre-print version: Loading...

      Authors: Tao Chen;Xiao Gao;Gang Liu;Chen Wang;Zeyang Zhao;Jie Dou;Ruiqing Niu;Antonio J. Plaza;
      Pages: 3648 - 3663
      Abstract: Landslides are catastrophic geological events that can cause significant damage to properties and result in the loss of human lives. Deep-learning technology applied to optical remote sensing images can enable effective landslide-prone area detection. However, conventional landslide detection (LD) models often employ complex structural designs to ensure detection accuracy. The complexity often hampers the detection speed, rendering these models inadequate for the swift emergency monitoring of landslides. To address these problems, we propose a new lightweight deep-learning-based framework, BisDeNet, for efficient LD. To improve the efficiency of the proposed BisDeNet, we replaced the context path in the original BiSeNet with DenseNet due to its strong feature extraction ability, few required parameters, and low model complexity. Two sites with different and representative landslide developments were selected as the study areas to verify the performance of our proposed BisDeNet. Additionally, we introduced landslide causative factors to enhance the sampling dataset. To evaluate the effectiveness of our approach, we compared the performance of our BisDeNet with the performances of three other BiSeNet-based methods and an advanced transformer-based model data-efficient image transformer (DeiT). Our experimental results indicate that the F1-scores of BisDeNet in the two study areas are 0.9006 and 0.8850, which are 26.22% and 1.86% higher than the scores of BiSeNet, respectively, but slightly lower than that of the DeiT model. Furthermore, our proposed BisDeNet requires the fewest number of parameters and the least memory out of the five models.
      PubDate: TUE, 09 JAN 2024 09:18:02 -04
      Issue No: Vol. 17, No. null (2024)
       
  • PTRSegNet: A Patch-to-Region Bottom–Up Pyramid Framework for the
           Semantic Segmentation of Large-Format Remote Sensing Images

    • Free pre-print version: Loading...

      Authors: Shiyan Pang;Yepeng Shi;Hanchun Hu;Lizhi Ye;Jia Chen;
      Pages: 3664 - 3673
      Abstract: Semantic segmentation is a basic task in the interpretation of remote sensing images. Mainstream deep-learning-based semantic segmentation algorithms typically process images with small sizes. However, remote sensing images typically involve large areas with buildings and water, which have weak textures. Because of the limited range of receptive fields, the semantic segmentation of such areas from small images may lead to problems, such as loss of spatial features and inaccurate boundary extraction. To address these problems, this article devises a patch-to-region framework for the semantic segmentation of large-format remote sensing images. This framework has a bottom–up pyramid structure, where the bottom layer is a small image patch, referred to as a “patch,” and the upper layer is a large image region, referred to as a “region.” The classical semantic segmentation network is first used to process small image patches to obtain pixel-by-pixel semantic features. Then, the pixel-by-pixel semantic features are sparsely reduced into patch-level semantic feature vectors, and the semantic feature vectors of the entire image region are processed through the contextual information extractor to extract the global semantic feature vectors. Subsequently, an information aggregation module is used to integrate the global semantic feature vectors and semantic features to obtain new semantic features with both global and local information. Finally, a lightweight decoding module is used to process the new semantic features to obtain the final semantic segmentation result. The developed framework is evaluated over three public datasets. The results of extensive experiments show that the framework can achieve more accurate and reliable semantic segmentation of high-resolution remote sensing images than state-of-the-art semantic segmentation algorithms. Moreover, ablation studies are performed to verify the effectiveness of each module of the framework.
      PubDate: THU, 11 JAN 2024 09:15:59 -04
      Issue No: Vol. 17, No. null (2024)
       
  • An Adaptive Signal Photon Detection Method Based on DBSCAN for
           Photon-Counting Laser Altimeter

    • Free pre-print version: Loading...

      Authors: Xiangfeng Liu;Zhenhua Wang;Wuzhong Yang;Shixian Chen;Fengxiang Wang;Xiaowei Chen;Weiming Xu;Rong Shu;
      Pages: 3674 - 3686
      Abstract: Photon-counting light detection and ranging is very sensitive to ambient interference, target features, and instrument performance, especially for long-distance detection of spaceborne laser altimeter and measurement of complex land-cover types with steep terrain. It is crucial to extract the signal photons on the ground surface from the collected photon point cloud (PPC). An adaptive signal photon detection method is presented in this article, which combines histogram statistics and boxplot analysis with density-based spatial clustering of applications with noise (DBSCAN), to denoise the PPC data with strong and weak noise obtained by ice, cloud, and land elevation satellite-2 laser altimeter. First, a coarse denoising with histogram of elevation is conducted on the raw PPC to reduce the calculation amount. Second, a fine denoising based on adaptive DBSCAN is used to extract the signal photons, where the key parameters of elliptic filter kernel are automatically determined according to the topographic data situation. We compared it with other methods, including local distance statistics (LDS), traditional and modified DBSCAN, traditional and modified ordering points to identify cluster structure (OPTICS), and ATL08 data. Some quantitative indicators, such as recall (R), precision (P), and F-score (F), are used to evaluate its performance. The results show that; 1) the adaptive DBSCAN has the best performance on preserving the vertical structural characteristics of ground objects, and 2) the adaptive DBSCAN in the mean R, P, and F of three land covers (i.e., mountain forest, urban, and water areas) can get up to the maximum are 0.9852, 0.9675, and 0.9761, respectively; followed by ATL08 data with 0.9773, 0.9412, and 0.9536, modified OPTICS with 0.9684, 0.9460, and 0.9586, and modified DBSCAN with 0.9613, 0.9474, and 0.9544; and then OPTICS with 0.9444, 0.9397, and 0.9378, and the DBSCAN with 0.9444, 0.9355, and 0.9554; the last one is LDS with 0.9382, 0.9261, and 0.9422. The proposed method provides an alternative approach for rapid and accurate processing of PPC on complex terrain.
      PubDate: WED, 10 JAN 2024 09:17:23 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Multiscale Convolutional Mask Network for Hyperspectral Unmixing

    • Free pre-print version: Loading...

      Authors: Mingming Xu;Jin Xu;Shanwei Liu;Hui Sheng;Zhiru Yang;
      Pages: 3687 - 3700
      Abstract: Deep learning has gained popularity in hyperspectral unmixing (HU) applications recently due to its powerful learning and data-fitting capabilities. As an unmixing baseline network, the autoencoder (AE) framework performs well in HU by automatically learning low-dimensional embeddings and reconstructing data. Nevertheless, there are spectral variability and nonlinear mixing problems in the highly mixed region of hyperspectral images, which can cause interference to structures using only AE. Therefore, inspired by the effectiveness of mask modeling, we propose a multiscale convolutional mask network (MsCM-Net) for HU with two new strategies. First, we propose a mixed region mask strategy suitable for the HU task, and a multiscale convolutional AE is adopted as the unmixing baseline network to apply the mask strategy, making the method more robust in solving ill-posed unmixing problems. In addition, a new initialization strategy is used in which vertex component analysis (VCA) is combined with density-based spatial clustering of applications with noise (DBSCAN) to mitigate the impact of outliers and noise on initialization. The proposed MsCM-Net performs more accurately than state-of-the-art methods by comparison experiments on one synthetic and three real hyperspectral data sets. The effectiveness of the mixed region mask strategy and DBSCAN-VCA initialization is also demonstrated by ablation experiments.
      PubDate: WED, 10 JAN 2024 09:17:23 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Enhanced Troposphere Tomography: Integration of GNSS and Remote Sensing
           Data With Optimal Vertical Constraints

    • Free pre-print version: Loading...

      Authors: Saeed Izanlou;Saeid Haji-Aghajany;Yazdan Amerian;
      Pages: 3701 - 3714
      Abstract: This article explores the enhancement of Global Navigation Satellite Systems (GNSS) tropospheric tomography by integrating remote sensing data and employing various vertical constraints. Wet refractivity modeling, critical for understanding atmospheric dynamics, has shown promising advancements. Leveraging tropospheric data from the Ocean and Land Color Instrument (OLCI), this research addresses the issue of empty voxels that impede GNSS-based tomography due to satellite and receiver geometries. Incorporating tropospheric data from remote sensing sensors mitigates empty voxels, enhancing retrieval accuracy for tropospheric water vapor. This study evaluates various vertical constraint functions in tropospheric tomography, presenting eight tomography schemes that utilize GNSS and OLCI data, highlighting their capacity to fill empty voxels without relying on empirical horizontal constraints. Results highlight the superiority of using OLCI observations in accuracy. Validation against radiosonde measurements and Weather Research and Forecasting model outputs affirms the reliability of this approach. Integrating OLCI observations with GNSS data reduces the average root mean square error by approximately 27%, with the Gaussian function exhibiting superior vertical constraint performance.
      PubDate: TUE, 16 JAN 2024 09:16:10 -04
      Issue No: Vol. 17, No. null (2024)
       
  • A Novel Spaceborne SAR Constellation Scheduling Algorithm for Sea Surface
           Moving Target Search Tasks

    • Free pre-print version: Loading...

      Authors: Dacheng Liu;Sheng Chang;Yunkai Deng;Zhihui He;Feng Wang;Zixuan Zhang;Chuanzhao Han;Chunrui Yu;
      Pages: 3715 - 3726
      Abstract: With the expanding scope of human activities in marine environments, the efficient detection and tracking of mobile targets on the ocean's surface have become increasingly crucial. Synthetic aperture radar (SAR) constellation can obtain ground observation data based on user requests and subject to visibility conditions. Now it is an indispensable tool in sea surface moving target search tasks. Satellite constellation resources are scarce and limited, and user demands are diverse. How to rationally dispatch satellite constellation resources to meet user needs to the maximum extent and improve the application efficiency of satellite resources is an urgent scientific problem that needs to be solved. This article mainly expounds two respects of work. First, modeling SAR constellation scheduling problem for sea surface moving target search tasks to establish the objective function. Second, a novel multistrategy discrete constrained differential evolution algorithm denoted as MSDCDE is proposed in the article. The proposed MSDCDE algorithm integrates cross strategy based on discrete variables, constraint handling techniques, population restart strategy, and left-shift local strategy, which can effectively avoid falling into local optimality, thereby achieving global optimality and improving search and rescue performances. Six sets of experiments, totaling 215 runs, have been conducted to validate the effectiveness of the proposed resolution process framework and the MSDCDE algorithm. The proposed method demonstrated an over 48.98% performance improvement compared with some state-of-the-art algorithms and significantly reduced task completion time.
      PubDate: FRI, 19 JAN 2024 09:16:55 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Multiscale Superpixel-Guided Weighted Graph Convolutional Network for
           Polarimetric SAR Image Classification

    • Free pre-print version: Loading...

      Authors: Ru Wang;Yinju Nie;Jie Geng;
      Pages: 3727 - 3741
      Abstract: Polarimetric synthetic aperture radar (PolSAR) has attracted more attentions because of its excellent observation ability, and PolSAR image classification has become one of the significant tasks in remote sensing interpretation. Various types and sizes of land cover objects lead to misclassification, especially in the boundaries of different categories. To solve these issues, a multiscale superpixel-guided weighted graph convolutional network (MSGWGCN) is proposed for classifying PolSAR images. In the proposed MSGWGCN, multiscale superpixel features are imported into the weighted graph convolutional network to obtain higher level representation, which can make full use of land cover object information in PolSAR images. Moreover, to fuse pixel-level features at different scales, a multiscale feature cascade fusion module is built, which plays an important role in preserving classification details. Experiments on three PolSAR datasets indicate that the proposed MSGWGCN performs better than other advanced methods on PolSAR classification task.
      PubDate: WED, 17 JAN 2024 09:16:24 -04
      Issue No: Vol. 17, No. null (2024)
       
  • An Improved Ultrahigh-Resolution Stepped-Frequency Spaceborne SAR Imaging
           Algorithm

    • Free pre-print version: Loading...

      Authors: Zhe Li;Zegang Ding;Tianyi Zhang;Linghao Li;Han Li;Zehua Dong;Pengnan Zheng;
      Pages: 3742 - 3754
      Abstract: Frequency stepping is a widely used technique for ultrahigh-resolution synthetic aperture radar (SAR). Although reducing the burden of hardware, this technique increases the complexity of imaging algorithms due to the intersubband time offsets and intersubband errors of delay, amplitude, and phase. To address the above problems, an improved ultrahigh-resolution stepped-frequency spaceborne SAR imaging algorithm is proposed in this article. By generating subband images individually, performing intersubband error estimation based on primary points, and then synthesizing the subband images in the imaging domain, the proposed algorithm effectively avoids the problem of time offsets and significantly improves intersubband error compensation accuracy benefiting from the high SNR in the imaging domain. Besides, considering the characteristics of nonideal factors in frequency-stepped SAR, a series of error compensation methods aiming at stop-and-go approximation, ionospheric error, and tropospheric delay are integrated to the proposed algorithm. The effectiveness of the proposed algorithms is verified via computer simulations, and real data experiments are also conducted based on both an X-band spaceborne SAR system, Taijing 4-01, and a Ka-band spaceborne SAR system, Luojia 2-01.
      PubDate: THU, 18 JAN 2024 09:16:22 -04
      Issue No: Vol. 17, No. null (2024)
       
  • A Lightweight Theory-Driven Network and Its Validation on Public Fully
           Polarized Ship Detection Dataset

    • Free pre-print version: Loading...

      Authors: Ziyuan Yang;Peng Zhang;Ning Wang;Tao Liu;
      Pages: 3755 - 3767
      Abstract: The utilization of polarimetric synthetic aperture radar (PolSAR) enables the preservation of more comprehensive target scattering characteristics compared to conventional SAR. Despite the significant advancements in deep learning (DL) technology for single polarimetric SAR ship detection, there remains a scarcity of DL research specifically focused on PolSAR ship detection. This article proposed a lightweight theory-driven network (LT-Net) for combining domain knowledge and DL techniques, which is suitable to detect small targets and reduce the computational complexity. The LT-Net incorporates convolutional block attention module, and employs differential downsampling during the process of high-dimensional feature extraction. The integration of domain knowledge lends each convolution step a distinct physical significance. Due to the limited availability of public datasets, this article presents a public fully polarized ship detection (FPSD) dataset for the first time, which contains 853 Pauli pseudocolor maps (in JPG format) and multilook complex data (in TIF format), with a total of 1714 ship targets from AIRSAR, UAVSAR, and RadarSAT-2. The FPSD encompasses a range of scenarios, with the majority of ship targets being small targets. Validation on FPSD shows that LT-Net exhibits significantly lower time complexity and space complexity compared to four popular target detection networks, Meanwhile, LT-Net achieves the best detection performance in the Pauli pseudocolor maps.
      PubDate: TUE, 16 JAN 2024 09:16:10 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Research on Modeling the Nonlinear Response Function of TDI CMOS Imaging
           System

    • Free pre-print version: Loading...

      Authors: Tan Gao;Liangliang Zheng;Xiaobin Wu;Haolin Ji;Biao Yang;Wei Xu;
      Pages: 3768 - 3779
      Abstract: The camera response function (CRF) establishes a numerical mapping between focal plane radiance, camera imaging parameters, and the intensity of output images. It plays a significant role in areas such as high dynamic range imaging and image processing. To establish an accurate response model for time delay integration (TDI) complementary metal–oxide–semiconductor (CMOS) imaging systems, this article proposes a radiometric calibration method for TDI CMOS imaging systems based on complex real-world scene images. The study begins by conducting an extensive analysis of the data link within the TDI CMOS imaging system, which serves as the foundation for establishing its a priori theoretical response model. Subsequently, the problem of solving the CRF model is transformed into an overdetermined equation established through the least square method. The optimal solution for this equation is obtained by singular value decomposition, which leads to the derivation of a 3-D response function for the imaging system. Finally, under consistent optical radiation conditions, radiation calibration experiments are performed on various targets using a self-developed TDI CMOS imaging system. The CRF is obtained based on the captured experimental image data. Furthermore, this article's approach is compared with the widely adopted linear fitting method commonly used within the respective field. The experimental results show that the visually perceived quality, structural similarity, mean grayscale, mean gradient, entropy, and standard deviation of images synthesized using the CRF proposed in this article are closer to those of actual captured images. The proposed method demonstrates higher accuracy and can provide a reliable basis for applications such as radiation response calibration of on-orbit spaceborne payloads, selection of imaging parameters, and multiexposure fusion of remote sensing images.
      PubDate: TUE, 16 JAN 2024 09:16:09 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Quantitative Analysis of SAR Image Geometric Distortion and Its
           Application in Deformation Rate Fusion Mapping

    • Free pre-print version: Loading...

      Authors: Zhengyang Wang;Xu Ma;Junhuan Peng;Mengyao Shi;Yun Peng;Yuhan Su;Zhaocheng Guo;Wenwen Wang;
      Pages: 3780 - 3790
      Abstract: Interferometry synthetic aperture radar (InSAR) technology has emerged as a powerful tool for the early identification and monitoring of geological hazards in the Loess Plateau. Since the SAR satellite adopts the range-azimuth two-dimensional side-looking imaging mode, the corresponding field area of the pixel will change with slope, aspect, and incident angle, especially in areas with severe changes in terrain and landforms, which causes serious area distortion, leading to abnormal deformation rate mapping. Considering the SAR imaging geometry, terrain slope and aspect conditions, this article proposes a quantitative analysis method for accurate geometric distortion (GD) based on pixel area and corresponding field area and constructs a weighted fusion method based on distortion values for InSAR deformation rate maps in overlapping regions of adjacent tracks. Taking Lingtai County, Gansu Province as the research area and sentinel-1 data as the research data, we utilize the SBAS method to process the sentinel-1 data and estimate the deformation rate, and then use the GD value as a weight factor to estimate the fusion rate of the overlapping area of adjacent tracks. The results demonstrate that the weighted fusion method based on distortion values can achieve a high-quality mosaic effect of deformation rate maps of adjacent tracks, effectively improve the GD influence in deformation rate mapping, and the identification of geological hazards.
      PubDate: WED, 17 JAN 2024 09:16:24 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Impacts of Spaceborne GNSS-R System Observations on Storm Surge
           

    • Free pre-print version: Loading...

      Authors: Younghun Kang;Ethan Kubatko;Mohammad M. Al-Khaldi;Joel T. Johnson;Suranjan Nepal;Aaron Sines;
      Pages: 3791 - 3798
      Abstract: This work assesses the impact of the Cyclone Global Navigation Satellite System (CYGNSS) (full delay-Doppler map) informed wind fields on storm surge simulations using the constellation's storm observations. In order to assess the impact of the CYGNSS-enhanced wind fields, storm surge simulations are performed with the ADvanced CIRCulation (ADCIRC) model and validation studies are performed with high water mark data provided by the U.S. Geological Survey. To provide context for the CYGNSS-based results, comparisons to storm surge predictions using Modern-Era Retrospective analysis for Research and Applications, Version 2,wind fields are also performed using the example of hurricane Harvey. In this initial assessment, it is observed that augmenting existing wind estimates with information provided by Global Navigation Satellite Systems Reflectometry systems, such as CYGNSS, has the potential of improving surge predictions relative to existing sources of wind information.
      PubDate: THU, 18 JAN 2024 09:16:22 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Deep Evidential Remote Sensing Landslide Image Classification With a New
           Divergence, Multiscale Saliency and an Improved Three-Branched Fusion

    • Free pre-print version: Loading...

      Authors: Jiaxu Zhang;Qi Cui;Xiaojian Ma;
      Pages: 3799 - 3820
      Abstract: Hitherto, image-level classification on remote sensing landslide images has been paid attention to, but the accuracy of traditional deep learning-based methods still have room for improvement. The evidence theory is found efficient to boost the accuracy of neural networks, however, the present study argues three challenges that hinder the lead-in of this theory in deep landslide image classification. Aiming at the three problems, this study makes three improvements. For the interpretability and decision-invariance losses of three previous divergences, we propose a belief Jensen–Renyi divergence with properties proven. To couple the evidence theory with deep remote sensing landslide image classification, a channelwise multiscale visual saliency fusion is developed. We additionally find that the channelwise fusion is capable to reduce false recognition of networks as compared with original RGB images. To avoid decision failures in evidence-theoretic fusion process, we design an interpretability improved three-branched fusion. Experiments on Bijie Landslide dataset corroborate the synergistic benefits of the three improvements, where the proposal is compared with state-of-the-art image classification backbone networks, remote sensing image scene classifiers, evidence fusion algorithms, and versatile evidence-theoretic deep learning classifiers. We also evaluated the new method with two sort of image degradation, as well as an actual scenario in Luding County, China, whose data is publicly available.
      PubDate: FRI, 19 JAN 2024 09:16:55 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Optimizing Rice Field Mapping in the Northern Region of China: An
           Asynchronous Flooding Signal and Object-Based Method

    • Free pre-print version: Loading...

      Authors: Long Li;Daoqin Zhou;Kai Liu;Tian Shi;Chou Xie;Shudong Wang;Hang Li;Guannan Dong;Xueke Li;
      Pages: 3821 - 3835
      Abstract: Accurate delineation of paddy fields holds importance in ensuring food security, efficient water resource management, and precise evaluation of greenhouse gas emissions. Here we propose an innovative approach, the asynchronous flooding and object-based (AF-OB) model, aimed at optimizing phenology-based paddy field mapping. The AF-OB model capitalizes on the asynchronous flooding phenomenon observed between paddy fields and nonpaddy fields, along with the seasonal variations in the normalized difference vegetation index. The simple noniterative clustering algorithm is integrated to mitigate the common issue of the “pretzel effect” encountered in paddy field mapping. Evaluation through independent samples yields compelling results, with the paddy field map generated by the AF-OB method achieving an overall accuracy of 94.28%. The paddy fields extracted using the AF-OB method exhibit alignment with statistical data, surpassing comparable algorithms relying on alternative land use products in terms of visual quality. Furthermore, the AF-OB model exhibits stability across time, space, and sensors, thus enhancing its applicability and robustness. The outputs of the AF-OB method offer reference data for informed agricultural production planning and the effective management of water resources.
      PubDate: TUE, 23 JAN 2024 09:18:03 -04
      Issue No: Vol. 17, No. null (2024)
       
  • SDTU-Net: Stepwise-Drop and Transformer-Based U-Net for Subject-Sensitive
           Hashing of HRRS Images

    • Free pre-print version: Loading...

      Authors: Kaimeng Ding;Shiping Chen;Yue Zeng;Yanan Liu;Bei Xu;Yingying Wang;
      Pages: 3836 - 3849
      Abstract: As a new integrity authentication technology, subject-sensitive hashing has the ability to achieve subject-sensitive authentication for high-resolution remote sensing (HRRS) images and can provide a security guarantee for their subsequent use. However, existing research on subject-sensitive hashing focuses on improving the structure of the deep neural network of the algorithm to improve the algorithm's performance, which makes it necessary to reconstruct the training dataset or modify the network structure in the face of different integrity authentication requirements. In this article, we delve into the impact of dropout on subject-sensitive hashing and propose a stepwise-drop mechanism to address the robustness and tampering-sensitivity requirements of subject-sensitive hashing. On this basis, a network named stepwise-drop and transformer-based U-net (SDTU-net) is proposed for subject-sensitive hashing of HRRS images. SDTU-net can use our proposed stepwise-drop mechanism to determine the drop rate of different network layers, which makes it possible to adjust the algorithm performance without changing network structure and training data. Experiments show that our SDTU-net based subject-sensitive hashing has better overall performance compared with existing algorithms, especially at medium and low thresholds. Our approach solves the problem that the existing algorithms cannot balance robustness and tamper sensitivity at low thresholds.
      PubDate: MON, 22 JAN 2024 09:21:31 -04
      Issue No: Vol. 17, No. null (2024)
       
  • On the Mitigation of Phase Bias in SAR Interferometry Applications: A New
           Model Based on NDWI

    • Free pre-print version: Loading...

      Authors: Nuno Cirne Mira;João Catalão;Giovanni Nico;
      Pages: 3850 - 3859
      Abstract: This article presents a study of the relationship among decorrelation phase in synthetic aperture radar (SAR) interferogram, soil moisture, and water content in vegetation with the aim of mitigating the contribution of decorrelation phase in SAR interferometry estimates of terrain displacements. A methodology for the mitigation of the phase bias based on the temporal variation of the vegetation water content is presented. Decorrelation phases are computed using time series of Sentinel-1 images and compared with in situ measurements of soil moisture. It is shown that soil moisture can partially explain the observed values of decorrelation phases pointing out the role of vegetation water content. A new model is proposed to compute the contribution of vegetation to the decorrelation phase based on the normalized difference water index (NDWI) index. The methodology is applied to all short temporal baseline interferograms obtained from the time series of Sentinel-1 SAR images, using the NDWI maps generated from Sentinel-2 multispectral images. The cumulative displacement is computed by integrating the short temporal baseline interferograms, corrected for the land cover and soil moisture changes. It is shown that the proposed methodology can reduce the variance of estimated cumulative displacement in areas covered by vegetation.
      PubDate: TUE, 23 JAN 2024 09:18:03 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Knowledge Distillation-Based Lightweight Change Detection in
           High-Resolution Remote Sensing Imagery for On-Board Processing

    • Free pre-print version: Loading...

      Authors: Guoqing Wang;Ning Zhang;Jue Wang;Wenchao Liu;Yizhuang Xie;He Chen;
      Pages: 3860 - 3877
      Abstract: Deep learning (DL) has been introduced to change detection (CD) due to its powerful feature representation and robust generalization abilities. However, the application of large DL models has high computational complexity and massive storage requirements for achieving good performance. For disaster emergency response and other applications with high timeliness requirements, it is difficult to deploy large DL models on spaceborne edge devices with limited resources to achieve on-board CD processing. To address this limitation, a novel CD based on knowledge distillation (CDKD) method that combines prototypical contrastive distillation and channel-spatial-normalized (CSN) distillation is proposed. PC distillation represents the feature distribution by calculating the differences between the similarities of pixel features and their positive and negative prototypes, and improves the student model's detection ability in changed regions that have similar features to the background by mimicking the relative feature distribution. CSN distillation combines two distillation paradigms, channel normalization and spatial normalization, and guides the student model to comprehensively learn the knowledge contained in the output probabilities of the teacher model to accurately identify changed regions with complex shapes. The effectiveness and reliability of the proposed CDKD method are verified on three public remote sensing CD datasets, and extensive experiments and analyses show that the proposed CDKD method can be used to train lightweight models with comparable performance to that of large models.
      PubDate: THU, 18 JAN 2024 09:16:22 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Conventional to Deep Ensemble Methods for Hyperspectral Image
           Classification: A Comprehensive Survey

    • Free pre-print version: Loading...

      Authors: Farhan Ullah;Irfan Ullah;Rehan Ullah Khan;Salabat Khan;Khalil Khan;Giovanni Pau;
      Pages: 3878 - 3916
      Abstract: Hyperspectral image classification (HSIC) has become a hot research topic. Hyperspectral imaging (HSI) has been widely used in a wide range of real-world application areas due to the in-depth spectral information stored within each pixel. Noticeably, the detailed features, i.e., a nonlinear correlation between the obtained spectral data and the correlating HSI data object, generate efficient classification results that are complex for traditional techniques. Deep learning (DL) has recently been validated as an influential feature extractor that efficiently identifies the nonlinear issues that have arisen in various computer vision challenges. This motivates using DL for HSIC, which shows promising results. This survey provides a brief description of DL for HSIC and compares cutting-edge methodologies in the field. We will first summarize the key challenges for HSIC, and then, we will discuss the superiority of DL and DL ensemble in addressing these issues. In this article, we divide state-of-the-art DL methodologies and DL with ensemble into spectral features, spatial features, and combined spatial–spectral features in order to comprehensively and critically evaluate the progress (future research directions as well) of such methodologies for HSIC. Furthermore, we will take into account that DL involves a substantial percentage of labeled training images, whereas obtaining such a number for HSI is time and cost consuming. As a result, this survey describes some methodologies for improving the classification performance of DL techniques, which can serve as future recommendations.
      PubDate: FRI, 12 JAN 2024 09:16:33 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Subspace Projection Attention Network for GPR Heterogeneous Clutter
           Removal

    • Free pre-print version: Loading...

      Authors: Yanjie Cao;Xiaopeng Yang;Conglong Guo;Dong Li;Peng Yin;Tian Lan;
      Pages: 3917 - 3926
      Abstract: Clutter removal in ground-penetrating radar (GPR) based on deep learning has been studied in recent years. However, existing methods are primarily designed for homogeneous background conditions and utilize only local spatial information via the convolution operation. In order to solve these issues, a subspace projection attention (SPA) network is proposed for GPR heterogeneous clutter removal in this article. First, a heterogeneous concrete dataset based on a numerical model with randomly placed aggregates is constructed, which incorporates the complex electromagnetic propagation process accurately to improve the effectiveness for heterogeneous clutter removal. In addition, the clutter basis learning neural network is designed by integrating the SPA module into the skip connection paths of U-Net architecture. By learning the subspace basis vectors adaptively, the SPA exploits both local and global spatial information to extract target features precisely. At the same time, the feature maps are projected to the target subspace to remove heterogeneous clutter features. Finally, the performance and effectiveness of proposed method are validated by simulations and experiments.
      PubDate: WED, 17 JAN 2024 09:16:24 -04
      Issue No: Vol. 17, No. null (2024)
       
  • A SAR Multiple RFI Suppression Method via Frobenius Norm and Iterative
           Matrix Decomposition

    • Free pre-print version: Loading...

      Authors: Qiang Guo;Yuhang Tian;Liangang Qi;Yani Wang;Daren Li;Mykola Kaliuzhnyi;
      Pages: 3927 - 3939
      Abstract: The scarcity and sharing nature of the electromagnetic spectrum present a significant challenge to the stable operation of synthetic aperture radar (SAR) systems, as they are susceptible to interference from other devices operating in the same frequency band, known as radio-frequency interference (RFI). In this article, we propose an effective semiparametric method for suppressing multiple RFI, named the iterative matrix decomposition algorithm based on the Frobenius norm (FIMD), we employ CUR decomposition and a soft-threshold algorithm to update low-rank and sparse matrices within an alternate projection framework. It is observed that there exist distinct distribution characteristics between interference points and strong scattering points in the echo domain. We propose a novel and effective signal protection method, which effectively mitigates the risk of strong scattering points being misidentified as interference signals and subsequently eliminated. In addition, we utilize random singular value decomposition instead of traditional singular value decomposition to enhance convergence speed of. Simulation results demonstrate that our proposed method exhibits superior suppression capability and robustness under varying-to-interference ratio conditions and it can be applied to L0-raw products and L1-SLC products.
      PubDate: TUE, 23 JAN 2024 09:18:08 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Securing Fast and High-Precision Localization for Shallow Underground
           

    • Free pre-print version: Loading...

      Authors: Dan Wu;Liming Wang;Jian Li;Meiyan Liang;Yunpeng Kang;Qi Jiao;
      Pages: 3940 - 3955
      Abstract: Shallow underground explosive source localization technology is a key technology in the field of underground space localization. The existing approaches mainly aim to improve the localization accuracy, but need to deploy enormous sensors in the monitoring area, and rely on a large number of backend workstations to solve. These methods have the defects of considerable calculation and high time cost, and are hard to satisfy the precise and real-time requirements of onsite testing, ultimately resulting in slow localization speed and accurate localization failure. Fortunately, emerging deep reinforcement learning can effectively solve the problem of slow search policy by modeling the source localization as a Markov decision process (MDP). Therefore, a curiosity-driven deep dueling double Q-learning network (C-D3QN) is subsequently proposed to solve the above MDP. The overestimation problem is solved by decoupling selection and evaluation of the bootstrap action, and the action difference is effectively increased by introducing the dueling network that separately represents state values and action advantages. Meanwhile, the exploration is jointly reinforced by an intrinsic reward outputted from the curiosity module and an extrinsic reward supplied by the environment, guaranteeing the convergence to global optimal. Finally, extensive simulation results based on the outfield experiment data show that compared with other algorithms, the proposed scheme can significantly improve exploration ability and learning speed as well as generalization and robustness. In addition, compared to the baseline algorithm deep Q-learning network, the C-D3QN algorithm can offer an improved localization accuracy as high as 99.62% and an increased localization speed of 66.23%.
      PubDate: WED, 10 JAN 2024 09:17:21 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Impact of Atmospheric Aerosols on the Accuracy of IMERG Precipitation
           Estimates Over Northern China

    • Free pre-print version: Loading...

      Authors: Xiaoying Li;Sungmin O;Na Wang;Lichen Liu;Yinzhou Huang;
      Pages: 3956 - 3970
      Abstract: Accurate satellite precipitation estimates are vital for understanding global and large-scale regional water cycles. Among the many factors influencing satellite precipitation data quality, the detection and accuracy of precipitation products at different atmospheric aerosol concentrations are not well studied. In this study, we investigated the impact of atmospheric aerosols on the accuracy of satellite precipitation products (IMERG) over North China by comparing performance metrics such as bias, normalized root mean squared error, probability of detection (POD), and false alarm ratio (FAR) under different atmospheric aerosol conditions. The results revealed that IMERG generally exhibits poorer detectability and quantification under pollution condition. Based on the error decomposition, the estimated errors in autumn and winter were dominated by false biases, which are mainly affected by atmospheric aerosols. At the sensor level, the FARs of both infrared (IR) and passive microwave sensors show escalating trends as pollutant concentrations increase. The POD of IR sensors is affected by pollution. Pollution has a significant impact on IR detection capability. Our findings suggest that atmospheric aerosols may impact the accuracy of IMERG precipitation estimates over Northern China and need to be taken into consideration in the IMERG retrieval process and data utilization.
      PubDate: FRI, 19 JAN 2024 09:16:55 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Assessing Coastal Heritage Sustainability: Crustal Deformation and
           Sea-Level Trends at the Qaitbay Citadel in Alexandria, Egypt

    • Free pre-print version: Loading...

      Authors: Mohamed I. Abdelaal;Min Bao;Mohamed Saleh;Soha Hassan;Jiao Guo;Mengdao Xing;
      Pages: 3971 - 3984
      Abstract: The UNESCO Agenda 2030 emphasizes the preservation of cultural heritage sites, focusing on coastal heritage preservation, which still poses substantial difficulties. While earlier studies have addressed the overall consequences of natural hazards along the Alexandria coastline, there is a gap in how they specifically affect coastal heritage sites, e.g., the Qaitbay citadel (our case study). This work seeks to bridge this gap by assessing the critical hazards faced by the Qaitbay citadel, including crustal deformation due to tectonic events, earthquakes, and Sea Level Rise (SLR) resulting from climate change. To comprehensively assess these challenges, a stack of Sentinel-1 SAR datasets (2017-2021) was processed using the Persistent Scatterer InSAR (PS-InSAR) technique to conduct spatial and temporal deformation variations of the citadel's site and its buildings. GPS measurements of Alexandria's (e.g., ALX2) station were correlated with InSAR results within the same period. Furthermore, satellite altimetry data from 1993–2021 covering the citadel and surroundings were processed to highlight long-term SLR trends. The findings indicate 1) A subsidence rate of $-$1 $\pm$ 0.2 mm/yr, associated with the citadel's foundational structures and identified through the PS-InSAR analysis, was caused by load-bearing effects and sand migration beneath the citadel, 2) a vertical displacement of $-$1.3$\pm$0.6 mm/yr obtained from ALX2-GPS station, consistent with the LOS velocity rate of PS-InSAR time series analysis at that specific location, 3) a SLR trend of +3.96 mm/yr, with notable peaks potentially related to episodes of Northern Ionian Gyre reversal, that could result in changes in water mass redistribution in the surrounding region.
      PubDate: FRI, 12 JAN 2024 09:16:33 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Integrating Spectral and Spatial Bilateral Pyramid Networks for
           Pansharpening

    • Free pre-print version: Loading...

      Authors: Yue Que;Hanqing Xiong;Xue Xia;Jie You;Yong Yang;
      Pages: 3985 - 3998
      Abstract: Multispectral (MS) image pansharpening is a significant technology for remote sensing image analysis, aiming to restore a high-resolution MS image by merging a high-resolution panchromatic image with a low-resolution MS image. While the convolutional neural networks (CNNs) have garnered considerable attention for their exceptional fusion capabilities in recent years, the existing CNN-based methods cannot effectively integrate spectral–spatial information. In this article, we introduce a novel MS pansharpening framework that integrates spectral and spatial networks in a bilateral pyramid form, allowing for the extraction of hierarchical spectral–spatial information. The proposed reduced residual dense (RRD) module serves as the fundamental building block of the spatial network. The RRD module gradually reduces the dimension of feature maps and employs the concatenation of RRD with global residual learning for comprehensive feature representation. In the spectral network, we present a cooperative attention fusion module to further enhance the correlation between spectral and spatial features. Through extensive experiments conducted on benchmarking simulated and real datasets, our proposed framework consistently outperforms state-of-the-art methods, demonstrating its effectiveness in MS image pansharpening applications.
      PubDate: MON, 22 JAN 2024 09:21:31 -04
      Issue No: Vol. 17, No. null (2024)
       
  • A Detection Method With Antiinterference for Infrared Maritime Small
           Target

    • Free pre-print version: Loading...

      Authors: Xun Zhang;Aiyu Wang;Yan Zheng;Suleman Mazhar;Yasheng Chang;
      Pages: 3999 - 4014
      Abstract: In this article, a novel infrared maritime small target detection method, called local dissimilarity measure with anti-interference based on global graph clustering (LDMGGC), is proposed. The Wasserstein distance is introduced to calculate the dissimilarity of gray level distribution between a central region and its neighborhoods. These dissimilarities construct the feature of a region. With this feature, detection for recalling all suspected targets is achieved. As the maritime interferences among suspected targets are able to be clustered, relaxing mutual k nearest neighbor graph is introduced in global graph clustering for filtering interferences. With this method, real targets are detected and maritime interferences are filtered out. Experiments are conducted on three maritime datasets and a nonmaritime dataset for comparison. On three datasets, the proposed method achieves the best Receiver Operating Characteristic curves and Area Under Curve (0.99529, 0.99945, 0.99573, and 0.9906) values, meaning that the proposed method has high detection probability and low false-alarm ratio. Target Hit Rate (98.04%, 97.96%, 100%, and 99.24%) and Intersection of Union (0.8170, 0.7542, 0.5824, 0.7707) on four datasets of the proposed method show it has a strong ability to suppress the interferences.
      PubDate: TUE, 23 JAN 2024 09:18:03 -04
      Issue No: Vol. 17, No. null (2024)
       
  • A Robust and Automatic Algorithm for TLS–ALS Point Cloud Registration in
           Forest Environments Based on Tree Locations

    • Free pre-print version: Loading...

      Authors: Fariborz Ghorbani;Yi-Chen Chen;Markus Hollaus;Norbert Pfeifer;
      Pages: 4015 - 4035
      Abstract: Fusing of terrestrial laser scanning (TLS) and airborne laser scanning (ALS) point cloud data has been recognized as an effective approach in forest studies. In this regard, co-registration of point clouds is considered one of the crucial steps in the integration process. Co-registering point clouds in forest environments faces various challenges, including unstable features, extensive occlusions, different viewpoints, and differences in point cloud densities. To address these intricate challenges, this study introduces an automated and robust method for co-registering TLS and ALS point clouds based on the correspondence of individual tree locations in forest environments. Initially, the positions of individual trees in both TLS and ALS data are extracted. Then, a filtering approach is applied to eliminate positions with low potential for corresponding matches in the TLS and ALS dataset. Since larger trees in the TLS data have a higher potential for corresponding matches in the ALS data, an iterative process is applied to identify correspondences between trees in both datasets. After estimating transformation parameters, the co-registration process is executed. The proposed method is applied on six datasets with varying forest complexities. The results demonstrate a high success rate up to 100% if the starting position of the TLS plots are located within ∼4 hectares (∼2000 trees). Additionally, the potential of the proposed method for co-registering TLS data with ALS data across different search areas and varying number of trees is evaluated in detail. The outcomes indicate that successful co-registration of TLS plot with 50 m diameter to ALS data is successful in the best case within a search radius of approximately 113 hectares (∼60,000 tree locations) and in the worst case for around 20 hectares (∼10,000 tree locations) depending on the forest complexity.
      PubDate: WED, 17 JAN 2024 09:16:24 -04
      Issue No: Vol. 17, No. null (2024)
       
  • ANED-Net: Adaptive Noise Estimation and Despeckling Network for SAR Image

    • Free pre-print version: Loading...

      Authors: Xu Wang;Yanxia Wu;Changting Shi;Ye Yuan;Xue Zhang;
      Pages: 4036 - 4051
      Abstract: Synthetic aperture radar (SAR) images are often affected by a type of multiplicative noise known as “speckle” due to their active imaging characteristics. This property complicates the processing and interpretation of SAR images. While deep learning techniques have demonstrated success in despeckling, many models are tailored to specific noise levels. This specificity can limit a model's ability to generalize to real SAR images with varying noise levels, potentially leading to oversmoothing or overfocusing on specific details. To address these challenges, we present the Adaptive Noise Estimation and Despeckling Network (ANED-Net). This network consists of a noise-level estimation phase and a noise-level-guided nonblind denoising phase. During the nonblind denoising phase, we develop a noise-feature-guided denoising network. This network integrates a hierarchical encoder–decoder denoising module based on the Transformer block (T-unet) and a denoising enhancement control block. Together, they skillfully capture both local and global dependencies inherent in SAR images, facilitating effective noise removal. Furthermore, we also introduce a deep-attention mechanism to counteract the attentional collapse observed when the Transformer is extended in depth, enhancing the network's feature extraction capability and strengthening the model's denoising performance. Extensive tests on synthetic and real images show that ANED-Net is robust to different noise scenarios. It effectively mitigates speckle noise even at unspecified levels and outperforms many established methods.
      PubDate: WED, 17 JAN 2024 09:16:24 -04
      Issue No: Vol. 17, No. null (2024)
       
  • LRSD-ADMM-NET: Simultaneous Super- Resolution Imaging and Target Detection
           for Forward-Looking Scanning Radar

    • Free pre-print version: Loading...

      Authors: Wenchao Li;Boyang Zhang;Kefeng Li;Jianyu Yang;Junjie Wu;Yin Zhang;Yulin Huang;
      Pages: 4052 - 4061
      Abstract: Forward-looking imaging and target detection are highly desirable in many military and civilian fields, such as search and rescue, sea surface surveillance, airport surveillance, and guidance. However, during the processing procedure, imaging and target detection are usually regarded as two independent parts, which means that the imaging result will directly affect the detection performance. In this article, the LRSD-ADMM-net is proposed to achieve simultaneous super-resolution imaging and target detection for forward-looking scanning radar. First, low-rank and sparse constraints as regularization norms are incorporated to establish objective function, and the alternating direction multiplier method is used to solve simultaneous super-resolution and target detection problems. Then, the solving process is expanded into a neural network, where the weight parameters of each level are obtained through adaptive learning. At last, experiments are conducted to verify the effectiveness of the proposed method.
      PubDate: FRI, 19 JAN 2024 09:16:55 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Assessing the Sensitivity of Semiempirical Models to Spectral Data Quality
           and Sensor Settings When Estimating Leaf Chlorophyll Content

    • Free pre-print version: Loading...

      Authors: Dong Li;Hengbiao Zheng;Xia Yao;Yan Zhu;Weixing Cao;Tao Cheng;
      Pages: 4062 - 4070
      Abstract: Leaf chlorophyll content (LCC) is an important indicator of nitrogen content, and therefore, the accurate monitoring of LCC will benefit agronomists in guiding fertilizer applications. Remote sensing techniques have been widely used to estimate LCC from canopy reflectance spectra. However, there is no sensor specific to the estimation of LCC from canopy reflectance spectra, and it is unclear whether LCC estimation is sensitive to the reflectance quality (such as the noise level) or sensor settings (such as spectral resolution). To help design a sensor specific to the estimation of LCC, this study calibrated a semiempirical model based on the leaf area index-insensitive chlorophyll index (LICI) and evaluated its sensitivity to reflectance quality and sensor settings using simulated, measured, and artificial datasets. Our results indicated that random Gaussian noise in reflectance has limited effects on LCC estimation when the level of Gaussian noise is less than 2%, while the negative systematic bias in reflectance has clear effects on the LCC estimation. The LCC estimation accuracy is nearly independent of the spectral resolution (
      PubDate: FRI, 19 JAN 2024 09:16:55 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Edge-Computing-Enabled Deep Learning Approach for Low-Light Satellite
           Image Enhancement

    • Free pre-print version: Loading...

      Authors: Trong-An Bui;Pei-Jun Lee;Chun-Sheng Liang;Pei-Hsiang Hsu;Shiuan-Hal Shiu;Chen-Kai Tsai;
      Pages: 4071 - 4083
      Abstract: Edge computing enables rapid data processing and decision-making on satellite payloads. Deploying deep learning-based techniques for low-light image enhancement improves early detection and tracking accuracy on satellite platforms, but it faces challenges due to limited computational resources. This article proposes an edge-computing-enabled inference model specifically designed onboard satellites. The proposed model follows an encoder–decoder architecture to generate the illumination map with low multiplication matrix complexity, 25.52 GMac of $1920 \times 1200$ image size. To reduce nanosatellite hardware consumption with a single-precision floating-point format, the edge-computing-enabled inference model proposes a quantized convolution that computes signed values. The proposed inference model is deployed on Arm Cortex-M3 microcontrollers onboard satellite payload (86.74 times faster than normal convolution model) but also has a similar quality with the low-light enhanced in full-precision computing of lightweight training model by using the peak signal-to-noise ratio (average of 28.94) and structural similarity index (average of 0.85) metrics.
      PubDate: TUE, 23 JAN 2024 09:18:03 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Analysis of Nighttime Light Changes and Trends in the 1-Year Anniversary
           of the Russia–Ukraine Conflict

    • Free pre-print version: Loading...

      Authors: Lin Wang;Henggang Lei;Hanqiu Xu;
      Pages: 4084 - 4099
      Abstract: The Russia–Ukraine conflict has persisted for over a year, posing challenges in assessing and verifying the extent of damage through on-site investigations. Nighttime light (NTL) remote sensing, an emerging approach for studying regional conflicts, can complement traditional methods. This article employs National Aeronautics and Space Administration's Black Marble products to reveal the response characteristics of NTL intensity at national and state scales during the first anniversary of the conflict (January 2022 to February 2023) in Ukraine. The article used the NTL ratio index to assess the relative intensity of NTL and month-on-month change rate, nighttime light change rate index (NLCRI), and the rate (R value) of linear regression analysis to depict spatiotemporal dynamics. In addition, Theil–Sen median trend analysis and Mann–Kendall tests were employed to analyze intensity trends, with a “dual-threshold method” to reduce extensive noise interference. The results showed: At the national scale, the conflict resulted in an 84.0% decrease in NTL across Ukraine. At the state scale, the most severe NTL decline occurred near the southwestern border and eastern conflict zone under Ukrainian government control, witnessing over 80% decline rates. The correlation of decreases in NLCRI and R values with population displacement, infrastructure damage, or curfew measures demonstrated that the concentration of refugees and electricity facility restoration led to increased NLCRI and R values. Overall, NTL reflects critical moments at the national scale and provides insights into military intentions and humanitarian measures at the state scale. Therefore, NTL can effectively serve as a tool for observation and assessment in military conflicts.
      PubDate: TUE, 23 JAN 2024 09:18:08 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Fusing Global and Local Information Network for Tassel Detection in UAV
           Imagery

    • Free pre-print version: Loading...

      Authors: Jianxiong Ye;Zhenghong Yu;
      Pages: 4100 - 4108
      Abstract: Unmanned aerial vehicles (UAVs), equipped with sensors, have made a significant impact in the field of agricultural analysis. Maize, being one of the most vital crops worldwide, is intricately linked to its yield and the growth of tassels. Leveraging UAV imagery for the automatic monitoring of maize tassels holds the potential to drive the development of intelligent maize cultivation. Current research methods, nevertheless, are limited and lack robustness. To address the challenge of tassel detection in UAV images, we propose an innovative network, termed FGLNet. This network models the backbone with a 16x down-sampling to retain richer pixel information and enhances performance by effectively fusing global and local information through weighted mechanisms. Moreover, the scarcity of tassel data presents a substantial constraint. In this article, we publicly release a new dataset, named the maize tassels detection and counting UAV (MTDC-UAV), featuring annotated bounding boxes, to advance research in the agricultural domain. Although tassel detection and counting in aerial images pose formidable challenges, our approach demonstrates remarkable accuracy in evaluations based on the MTDC-UAV dataset. It achieves a detection AP$_{50}$ of 0.837 and a counting R$_{2}$ of 0.9409, all while maintaining a parameter count of just 0.77 M. This level of performance considerably outperforms other state-of-the-art computer vision methods. Overall, this research not only introduces innovative concepts but also provides worthwhile references and a solid data foundation for future studies.
      PubDate: MON, 22 JAN 2024 09:21:31 -04
      Issue No: Vol. 17, No. null (2024)
       
  • XGBoost-Based Analysis of the Relationship Between Urban 2-D/3-D
           Morphology and Seasonal Gradient Land Surface Temperature

    • Free pre-print version: Loading...

      Authors: Xinyue Ma;Jun Yang;Rui Zhang;Wenbo Yu;Jiayi Ren;Xiangming Xiao;Jianhong Xia;
      Pages: 4109 - 4124
      Abstract: The escalation of greenhouse gas emissions has led to a continuous rise in land surface temperature (LST). Studies have highlighted the substantial influence of urban morphology on LST; however, the impact of different dimensional indicators and their gradient effects remain unexplored. Selecting the urban area of Shenyang as a case, we chose various indicators representing different dimensions. By employing XGBoost for regression analysis, we aimed to explore the effects of urban 2-D and 3-D morphology on seasonal LST and its gradient effect. The following results were obtained: 1) the spatial pattern of LST in spring and winter in Shenyang was higher in the suburbs than in the center; 2) the correlation patterns of the indicators in spring and winter were similar, except for the proportion of woodland and grass, digital elevation model, and sky view factor, which exhibited opposing trends in summer and autumn; 3) vegetation and construction had the highest influence on LST in the 2-D index, followed by building forms and natural landscapes in the 3-D urban morphology; and 4) the influence of each indicator varied significantly across different gradients. Among all the indicators, the landscape index, social development, building forms, and skyscape had the highest impacts on urban areas. Vegetation and built-up areas had a greater influence on suburban areas. The findings of this study can assist in adjusting urban morphology and provide valuable recommendations for targeted improvements in thermal environments, thereby contributing to urban sustainable development.
      PubDate: MON, 01 JAN 2024 09:18:10 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Deep Content-Dependent 3-D Convolutional Sparse Coding for Hyperspectral
           Image Denoising

    • Free pre-print version: Loading...

      Authors: Haitao Yin;Hao Chen;
      Pages: 4125 - 4138
      Abstract: Despite the significant successes in hyperspectral image (HSI) denoising, pure data-driven HSI denoising networks still suffer from limited understanding of inference. Deep unfolding (DU) is a feasible way to improve the interpretability of deep network. However, the specialized spatial-spectral DU methods are seldom studied, and the simple spatial-spectral extension leads to unpleasant spectral distortion. To tackle these issues, we first propose a content-dependent 3-D convolutional sparse coding (CD-CSC) to jointly represent spatial-spectral feature. Specifically, the 3-D filters used in CD-CSC for each HSI are unique, which are determined by linear combination of base 3-D filters. Then, we develop a novel CD-CSC-inspired DU network for HSI denoising, called CD-CSCNet. Furthermore, by exploiting the lightweight of separable convolution and the adaptability of hypernetwork, we design a separable content-dependent 3D Convolution (SCD-Conv) to carry out CD-CSCNet. SCD-Conv not only reduces computational complexity, but also can be viewed as the convolutional sparse coding with spatial and spectral dictionaries. Extensive experimental results on the ICVL, Zhuhai-1 OHS-3C, and GaoFen-5 datasets demonstrate that CD-CSCNet outperforms several recent pure data-driven and DU-based networks quantitatively and visually.
      PubDate: TUE, 23 JAN 2024 09:18:08 -04
      Issue No: Vol. 17, No. null (2024)
       
  • The Displacement Analysis and Prediction of a Creeping Ancient Landslide
           at Suoertou, Zhouqu County, China

    • Free pre-print version: Loading...

      Authors: Yumin Fang;Lifeng Zhang;Yi He;Wang Yang;Tianbao Huo;Qing Zhang;Jiangang Lu;
      Pages: 4139 - 4163
      Abstract: The ancient Suoertou landslide seriously threatens the surrounding population's lives and property. Monitoring and predicting this landslide is crucial to ensure the affected areas’ safety. The previous research on the landslide's displacement characteristics and mechanisms has lacked detailed analyses. In addition, its future development trends must be understood. Therefore, we conducted a detailed analysis of the ancient Suoertou landslide's displacement characteristics and mechanisms using small baseline subset interferometric synthetic aperture radar monitoring results from 2018 to 2023. Furthermore, by applying a gated recurrent unit prediction mode that incorporates the refined displacement characteristics and mechanisms, we forecasted the landslide's displacement trends. The results show that this landslide is currently undergoing overall slow displacement with violent fluctuations in localized areas. Tectonic movement, precipitation, human activities, river erosion, and other factors interact, forming a vicious development displacement mechanism. According to our prediction, the displacement of this landslide will be in a trend of fluctuating increase from June 2023 to June 2024. In particular, the local area will undergo abnormal displacement acceleration. The results of the current research provide a scientific basis upon which to monitor landslides, promote their management, and reduce the risk of losses due to landslide disasters.
      PubDate: TUE, 23 JAN 2024 09:18:08 -04
      Issue No: Vol. 17, No. null (2024)
       
  • DSHNet: A Semantic Segmentation Model of Remote Sensing Images Based on
           Dual Stream Hybrid Network

    • Free pre-print version: Loading...

      Authors: Yujia Fu;Xiangrong Zhang;Mingyang Wang;
      Pages: 4164 - 4175
      Abstract: Semantic segmentation is an important issue in intelligent interpretation of remote sensing, playing an important role in applications such as Earth observation and land data update. However, remote sensing images often contain complex ground objects and the boundaries between them are blurred, which poses a huge challenge to the semantic segmentation task of remote sensing images. This article proposes a dual stream hybrid network (DSHNet) model, which focuses on parallel extraction of semantic and boundary features in remote sensing images, and improves the performance of semantic segmentation by fully integrating dual stream information. In the semantic stream, the ViT model pretrained on remote sensing images is used as the backbone network for feature extraction. In the boundary stream, the boundary detection operator Sobel is used to capture the boundaries of different ground objects in the image, and a boundary enhancement mechanism is taken to optimize and enhance the feature representation of ground object boundaries. In addition, DSHNet designs a feature fusion module to cross-aggregate features from both semantic and boundary streams. Compared with the state-to-art semantic segmentation methods, DSHNet model has achieved the best performance on two datasets of Yellow River Estuary Wetland and Gaofen image dataset.
      PubDate: MON, 05 FEB 2024 09:16:30 -04
      Issue No: Vol. 17, No. null (2024)
       
  • DHRNet: A Dual-Branch Hybrid Reinforcement Network for Semantic
           Segmentation of Remote Sensing Images

    • Free pre-print version: Loading...

      Authors: Qinyan Bai;Xiaobo Luo;Yaxu Wang;Tengfei Wei;
      Pages: 4176 - 4193
      Abstract: In the field of remote sensing image processing, semantic segmentation has always been a hot research topic. Currently, deep convolutional neural networks (DCNNs) are the mainstream methods for the semantic segmentation of remote sensing image (RSI). There are two commonly used semantic segmentation methods based on DCNNs: multiscale feature extraction based on deep-level features, and global modeling. The former can better extract object features of different scales in complex scenes. However, this method lacks sufficient spatial information, resulting in poor edge segmentation ability. The latter can effectively solve the problem of limited receptive field in DCNNs obtaining more comprehensive feature extraction results. Unfortunately, this method is prone to misclassification, resulting in incorrect predictions of local pixels. To address these issues, we propose the dual-branch hybrid reinforcement network (DHRNet) for more precise semantic segmentation of RSI. This model is a dual-branch parallel structure with a multiscale feature extraction branch and a global context and detail enhancement branch. This structure decomposes the complex semantic segmentation task, allowing each branch to extract features with different emphases while retaining sufficient spatial information. The results of both branches are fused to obtain a more comprehensive segmentation result. After conducting extensive experiments on three publicly available RSI datasets, ISPRS Potsdam, ISPRS Vaihingen, and LoveDA, DHRNet demonstrates excellent results with the mean intersection over union of 86.97%, 83.53%, and 54.48% on the three datasets, respectively.
      PubDate: TUE, 23 JAN 2024 09:18:03 -04
      Issue No: Vol. 17, No. null (2024)
       
  • 3-D High-Resolution ISAR Imaging for Noncooperative Air Targets

    • Free pre-print version: Loading...

      Authors: Marcin Kamil Baczyk;Piotr Samczynski;Jedrzej Drozdowicz;Maciej Wielgo;Jakub Sobolewski;Marek Ciesielski;Jakub Julczyk;Krzysztof Stasiak;Grzegorz Pietrzykowski;Karol Abratkiewicz;Maciej Soszka;
      Pages: 4194 - 4207
      Abstract: This article uses the inverse synthetic aperture radar (ISAR) imaging method to present real-world tests on 3-D radar imaging of noncooperative air targets. Initially, the fundamentals of 3-D ISAR are introduced. This is followed by a discussing of the challenges of obtaining high-quality 3-D radar images. An essential feature of the applied method is its basis on the back-projection family of techniques, eliminating the need for iterative image reconstruction. These theoretical concepts are validated using both simulations and real-life signals. This article also provides insights into the measurement campaign and the signal processing techniques applied to achieve the presented results.
      PubDate: TUE, 23 JAN 2024 09:18:03 -04
      Issue No: Vol. 17, No. null (2024)
       
  • AFE-Net: Attention-Guided Feature Enhancement Network for Infrared Small
           Target Detection

    • Free pre-print version: Loading...

      Authors: Keyan Wang;Xueyan Wu;Peicheng Zhou;Zuntian Chen;Rui Zhang;Liyun Yang;Yunsong Li;
      Pages: 4208 - 4221
      Abstract: Infrared small target detection is considerably challenging due to the few pixels in targets, low signal-to-noise ratio, and complex background. In this article, we propose an effective attention-guided feature enhancement network (AFE-Net), which can leverage the local and nonlocal features of targets and background in infrared images. The AFE-Net consists of three key modules, namely encoder and decoder interactive guidance (EDIG) module, cascading false alarm removal (CFAR) module, and random scale input (RSI) module. Specifically, in the EDIG module, we employ a CA mechanism on encoding and decoding layers to select feature channels with higher contribution. Then, we impose a bottom-up pointwise attention block to highlight the features of small infrared targets and suppress possible noise by incorporating the low-level detailed features into the high-level semantic features. The CFAR module extracts affluent global features by cascading nonlocal operations of different layers, which can remove clutters with similar features to infrared targets. The RSI module is placed in front of the entire detection network to extract multiscale features of infrared small targets, which can enhance the robustness of the proposed network. Experimental results on the SIRST dataset and comprehensive comparisons with representative methods demonstrate the superiority of our proposed method.
      PubDate: TUE, 16 JAN 2024 09:16:09 -04
      Issue No: Vol. 17, No. null (2024)
       
  • GFFNet: Global Feature Fusion Network for Semantic Segmentation of
           Large-Scale Remote Sensing Images

    • Free pre-print version: Loading...

      Authors: Yong Cao;Chunlei Huo;Shiming Xiang;Chunhong Pan;
      Pages: 4222 - 4234
      Abstract: Semantic segmentation plays a pivotal role in interpreting high-resolution remote sensing images (RSIs), where contextual information is essential for achieving accurate segmentation. Despite the common practice of partitioning large RSIs into smaller patches for deep model input, existing methods often rely on adaptations from natural image semantic segmentation techniques, limiting their contextual scope to individual images. To address this limitation and harness a broader range of contextual information from original large-scale RSIs, this study introduces a global feature fusion network (GFFNet). GFFNet employs a novel approach by incorporating a group transformer structure alternated with group convolution, forming a lightweight global context learning branch. This design facilitates the extraction of global contextual features from the large-scale RSIs. In addition, we propose a cross feature fusion module that seamlessly integrates local features obtained from the convolutional network with the global contextual features. GFFNet serves as a versatile plugin for existing RSI semantic segmentation models, particularly beneficial when the target dataset involves cropping. This integration enhances the model's performance, especially in terms of segmenting large-scale objects. Experimental results on the ISPRS and GID-15 datasets validate the effectiveness of GFFNet in improving segmentation capabilities for large-scale objects in RSIs.
      PubDate: MON, 29 JAN 2024 09:16:57 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Toward the Next Generation of Microwave Sounders: Benefits of a Low-Earth
           Orbit Hyperspectral Microwave Instrument in All-Weather Conditions Using
           AI

    • Free pre-print version: Loading...

      Authors: Eric S. Maddy;Flavio Iturbide-Sanchez;Sid Ahmed Boukabara;
      Pages: 4235 - 4246
      Abstract: This study presents scientific results that serve as arguments for advocating the development of a hyperspectral microwave sensor (HyMS). Through simulation experiments, the results of this study demonstrate the major benefits of HyMS sensor observations in low-Earth orbit (LEO), including; 1) increased information content over the microwave region, 2) improved temperature and moisture sounding in all-weather conditions, resulting from higher signal-to-noise ratios, finer vertical resolution, and a reduced dependence on background information due to the increased spectral resolution around oxygen and water vapor absorption features between 23 and 183 GHz, 3) improved profiling of hydrometeors, and 4) improved resilience to radio frequency interference, demonstrated at 23 GHz, associated with the redundant information provided by the HyMS. The deployment of HyMS instruments in LEO orbit is expected to provide an improved knowledge of the state of the atmosphere, particularly if deployed in the form of a constellation, due to the enhanced temporal, spatial, and spectral resolution capabilities that those sensors can provide with respect to present meteorological microwave sounders. This work takes advantage of artificial intelligence (AI), particularly its capability to rapidly and simultaneously process hundreds of channels and retrieve large sets of geophysical parameters, to assess the impact of HyMS in geophysical space. The results presented in this manuscript are expected to contribute to the design of the next generation of microwave sounders, but also to consider the usage of AI to fully exploit the information content provided by these sensors, particularly if deployed in the form of a constellation of satellites.
      PubDate: MON, 22 JAN 2024 09:21:31 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Superpixel-level CFAR Ship Detection Based on Polarimetric Bilateral
           Truncated Statistics

    • Free pre-print version: Loading...

      Authors: Wenxing Mu;Ning Wang;Lu Fang;Tao Liu;
      Pages: 4247 - 4262
      Abstract: Constant false alarm rate (CFAR) detector is a common method for ship detection in polarimetric synthetic aperture radar (PolSAR) images. CFAR detectors greatly depend on the clutter modeling that can be easily affected by the contamination caused by both lower- and higher-intensity outliers, such as spilled oil and intensive targets. Traditional CFAR detectors perform detection in a pixel-by-pixel manner, which ignores the spatial information. Both the bias in clutter modeling and the absence of spatial information can degrade the ship target detection performance. In this study, a superpixel-level polarimetric bilateral truncated statistics CFAR detector is proposed to promote the ship target detection performance in complex ocean scenarios. As the preprocessing of the PolSAR image, the superpixel segmentation is conducted based on the multilook polarimetric whitening filter result to select candidate ship target superpixels for bilateral truncation and background clutter modeling. The elliptical truncation is expanded to a complex situation and the relationship between the second moments before and after truncation is derived. The maximum-likelihood estimation estimator of the equivalent number of looks based on the bilateral truncation distribution is derived and compared with other parameter estimators. The influence of the truncation depth on estimator performance is analyzed, according to which the adaptive bilateral truncation method is determined. The Gaussian mixture model and the Parzen window kernel method are compared with the model-based method and utilized for data fitting. The proposed method performs bilateral truncation based on the superpixel segmentation result to provide pure clutter samples for accurate parameter estimation and clutter distribution modeling, reducing time consumption and false alarms. The method is validated efficient on both simulated and measured data from RADARSAT-2.
      PubDate: MON, 22 JAN 2024 09:21:31 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Improving ATMS Imagery Visualization Using Limb Correction and AI
           Resolution Enhancement

    • Free pre-print version: Loading...

      Authors: Xingming Liang;Lihang Zhou;Mitch Goldberg;Satya Kalluri;Christopher Grassotti;Ninghai Sun;Banghua Yan;Hu Yang;Lin Lin;Quanhua Liu;
      Pages: 4263 - 4279
      Abstract: The advanced technology microwave sounder (ATMS) is an important satellite instrument that provides vital data on atmosphere temperature and moisture for weather forecasting and climate research, and helps us plan for extreme weather. However, its coarse resolution and angular dependence have long been a challenge for improving image visualization. This article proposes a method to enhance the imagery visualization for ATMS, combining limb correction (LC) with artificial intelligence (AI) resolution enhancement (RE). Measurement data from the ATMS onboard NOAA-20 were utilized to train the LC method, which were then validated using newly acquired NOAA-21 ATMS data. The AI RE was performed using enhanced super-resolution generative adversarial networks, which increased the pixel resolution by a factor of four. The high-resolution (HR) Advanced Microwave Scanning Radiometer 2 data served as a reference to initially and quantitatively evaluate the RE method. The combined method of LC and AI RE produced an angular-dependence-free and HR ATMS image, resulting in a significant improvement in image visualization, including surface and atmosphere information, and allows for clear identification of severe weather events. For the swift identification and analysis of tropical cyclones in the upcoming season, as of this writing, this proposed method has been routinely employed to produce high-quality global ATMS images, and these images are showcased and tested in the NOAA internal HR imagery visualization system—JSTAR Mapper. Moreover, concentrated efforts are being made to further enhance these images in preparation for an official release.
      PubDate: THU, 18 JAN 2024 09:16:22 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Long-Term Prediction of Sea Surface Temperature by Temporal Embedding
           Transformer With Attention Distilling and Partial Stacked Connection

    • Free pre-print version: Loading...

      Authors: Hao Dai;Zhigang He;Guomei Wei;Famei Lei;Xining Zhang;Weijie Zhang;Shaoping Shang;
      Pages: 4280 - 4293
      Abstract: Sea surface temperature (SST) is one of the most important parameters in the global ocean–atmosphere system, and its long-term changes will have a significant impact on global climate and ecosystems. Accurate prediction of SST, therefore, especially the improvement of long-term predictive skills is of great significance for fishery farming, marine ecological protection, and planning of maritime activities. Since the effective and precise description of the long-range dependence between input and output requires higher model prediction ability, it is an extremely challenging task to achieve accurate long-term prediction of SST. Inspired by the successful application of the transformer and its variants in natural language processing similar to time-series prediction, we introduce it to the SST prediction in the China Sea. The model Transformer with temporal embedding, attention Distilling, and Stacked connection in Part (TransDtSt-Part) is developed by embedding the temporal information in the classic transformer, combining attention distillation and partial stacked connection, and performing generative decoding. High-resolution satellite-derived data from the National Oceanic and Atmospheric Administration is utilized, and long-term SST predictions with day granularity are achieved under univariate and multivariate patterns. With root mean square error and mean absolute error as metrics, the TransDtSt-Part outperforms all competitive baselines in five oceans (i.e., subareas of Bohai, Yellow Sea, East China Sea, Taiwan Strait, and South China Sea) and six prediction horizons (i.e., 30, 60, 90, 180, 270, and 360 days). Experimental results demonstrate that the performance of the innovative model is encouraging and promising for the long-term prediction of SST.
      PubDate: THU, 25 JAN 2024 09:16:31 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Large Kernel Separable Mixed ConvNet for Remote Sensing Scene
           Classification

    • Free pre-print version: Loading...

      Authors: Keqian Zhang;Tengfei Cui;Wei Wu;Xueke Zheng;Gang Cheng;
      Pages: 4294 - 4303
      Abstract: Among tasks related to intelligent interpretation of remote sensing data, scene classification mainly focuses on the holistic information of the entire scene. Compared with pixel-level or object-based tasks, it involves a richer semantic context, making it more challenging. With the rapid advancement of deep learning, convolutional neural networks (CNNs) have found widespread applications across various domains, and some work has introduced them into scene classification tasks. However, traditional convolution operations involve sliding small convolutional kernels across an image, primarily focusing on local details within a small receptive field. To achieve better modeling of the entire image, the smaller receptive field limits the ability of convolution operation to capture features over a broader range. To this end, we introduce large kernel CNNs into the scene classification task to expand the receptive field of the mode, which allows us to capture comprehensive nonlocal information while still acquiring rich local details. However, in addition to encoding spatial association, the effective information within the feature maps is also strongly channel related. Therefore, to fully model this channel dependency, a novel channel separation and mixing module has been designed to realize feature correlation in the channel dimension. The combination of them forms a large kernel separable mixed ConvNet, enabling the model to capture effective dependencies of feature maps in both spatial and channel dimensions, thus achieving enhanced feature expression. Extensive experiments conducted on three datasets have also validated the effectiveness of the proposed method.
      PubDate: MON, 15 JAN 2024 09:20:19 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Structures Detection Based on CLSK Model Combined With Shadow Information
           Using High-Resolution Remote Sensing Images

    • Free pre-print version: Loading...

      Authors: Chengrui Wang;Kai Xu;Rong Li;Zhiyong Zhang;Kun Qin;Yubin Xu;
      Pages: 4304 - 4319
      Abstract: In light of rapid economic and urban growth, the proliferation of structures including transmission towers, signal poles, and wind generators has become evident. Consequently, precise object detection of these structures emerges as a pivotal approach to enhance infrastructure management. This technique establishes a robust foundation for achieving elevated efficiency, precision, optimized energy management, and heightened safety monitoring. This article introduces a novel model for detecting structure based on channel and large-scale selective kernel (CLSK) model using high-resolution images. The method is rooted in a two-stage target detection network, enabling simultaneous identification of both primary structures and their associated shadows. The incorporation of deformable convolutions augments the model's ability to extract intricate features. Moreover, the introduction of the innovative LSK attention module, along with the CLSK attention module, enhances the optimization of features gleaned from the network's core architecture. Simultaneously, the complete intersection over union loss function refines the network's focus by considering parameters such as center point distance and aspect ratio in addition to the conventional overlapping area. This comprehensive approach facilitates improved feedback on detection outcomes. Empirical evaluation of the proposed network underscores its superior performance when juxtaposed with both conventional network models and the rotating detection box network.
      PubDate: FRI, 19 JAN 2024 09:16:55 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Marine Debris Detection in Satellite Surveillance Using Attention
           Mechanisms

    • Free pre-print version: Loading...

      Authors: Ao Shen;Yijie Zhu;Plamen Angelov;Richard Jiang;
      Pages: 4320 - 4330
      Abstract: Marine debris poses a critical threat to environmental ecosystems, necessitating effective methods for its detection and localization. This study addresses the existing limitations in the literature by proposing an innovative approach that combines the instance segmentation capabilities of YOLOv7 with various attention mechanisms to enhance efficiency and broaden applicability. The primary contribution lies in the exploration and comparison of three attentional models: lightweight coordinate attention, combining spatial and channel focus (CBAM), and bottleneck transformer based on self-attention. Leveraging a meticulously labeled dataset of satellite images containing ocean debris, the study conducts a comprehensive assessment of box detection and mask evaluation. The results demonstrate that CBAM emerges as the standout performer, achieving the highest F1 score (77%) in box detection, surpassing coordinate attention (71%) and YOLOv7/bottleneck transformer (both around 66%). In mask evaluation, CBAM continues to lead with an F1 score of 73%, while coordinate attention and YOLOv7 exhibit comparable performances (around F1 scores of 68% and 69%), and bottleneck transformer lags behind at an F1 score of 56%. This compelling evidence underscores CBAM's superior suitability for detecting marine debris compared to existing methods. Notably, the study reveals an intriguing aspect of the bottleneck transformer, which, despite lower overall performance, successfully detected areas overlooked by manual annotation. Moreover, it demonstrated enhanced mask precision for larger debris pieces, hinting at potentially superior practical performance in certain scenarios. This nuanced finding underscores the importance of considering specific application requirements when selecting a detection model, as the bottleneck transformer may offer unique advantages in certain contexts.
      PubDate: WED, 03 JAN 2024 09:17:51 -04
      Issue No: Vol. 17, No. null (2024)
       
  • A Joint Architecture of Mixed-Attention Transformer and Octave Module for
           Hyperspectral Image Denoising

    • Free pre-print version: Loading...

      Authors: Mahmood Ashraf;Lihui Chen;Xichuan Zhou;Muhammad Allah Rakha;
      Pages: 4331 - 4349
      Abstract: Convolutional neural networks (CNNs) recently have achieved impressive performance for hyperspectral image denoising. However, current CNNs have limitations in exploring spectral correlations across various bands and the interactions among features within each band. Although transformers are introduced to capture spatial-spectral correlation in hyperspectral image (HSI), generally, they either explore intercorrelation in bands or intracorrelation between bands, neglecting the combination of intra- and intercorrelation in HSI cubes. Besides, transformer methods rarely address hierarchical (i.e., the low- and high-level) features in an adaptive manner. That is, features at different levels are of different importance, whereas these features are tackled equally in these methods. To alleviate these limitations, we introduce a joint architecture (the so-called MAOTformer) of mixed-attention transformers and octave modules for HSI denoising. On the one hand, the mixed attention transformer (MATB) is designed to simultaneously capture pixel relationships inter- and intrabands by incorporating naive spatial self-attention, bidirectional recurrent channel attention, and progressive channel attention. Besides, a U-net based on mixed attention is equipped with attentive skip connections in MATB, which enables the proposed MAOTformer to explore hierarchical features by U-net and adaptively connect these hierarchical features by the attentive skip connections. On the other hand, we introduce an octave module behind each MATB to utilize multiscale features for separating noise in high-frequency components. Extensive experiments are conducted on synthetic and real-world HSIs, showing that the proposed method outperforms state-of-the-art methods.
      PubDate: MON, 22 JAN 2024 09:21:31 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Self-Supervised Spatio-Temporal Representation Learning of Satellite Image
           Time Series

    • Free pre-print version: Loading...

      Authors: Iris Dumeur;Silvia Valero;Jordi Inglada;
      Pages: 4350 - 4367
      Abstract: In this article, a new self-supervised strategy for learning meaningful representations of complex optical satellite image time series (SITS) is presented. The methodology proposed, named Unet-BERT spAtio-temporal Representation eNcoder (U-BARN), exploits irregularly sampled SITS. The designed architecture allows learning rich and discriminative features from unlabeled data, enhancing the synergy between the spatio-spectral and the temporal dimensions. To train on unlabeled data, a time-series reconstruction pretext task inspired by the BERT strategy but adapted to SITS is proposed. A Sentinel-2 large-scale unlabeled dataset is used to pretrain U-BARN. During the pretraining, U-BARN processes annual time series composed of a maximum of 100 dates. To demonstrate its feature learning capability, representations of SITS encoded by U-BARN are then fed into a shallow classifier to generate semantic segmentation maps. Experimental results are conducted on a labeled crop dataset (PASTIS) as well as a dense land cover dataset (MultiSenGE). Two ways of exploiting U-BARN pretraining are considered: either U-BARN weights are frozen or fine-tuned. The obtained results demonstrate that representations of SITS given by the frozen U-BARN are more efficient for land cover and crop classification than those of a supervised-trained linear layer. Then, we observe that fine-tuning boosts U-BARN performances on MultiSenGE dataset. In addition, we observe on PASTIS, in scenarios with scarce reference data that the fine-tuning brings a significative performance gain compared to fully supervised approaches. We also investigate the influence of the percentage of elements masked during pretraining on the quality of the SITS representation. Eventually, semantic segmentation performances show that the fully supervised U-BARN architecture reaches better performances than the spatio-temporal baseline (U-TAE) on both downstream tasks: crop and dense land cover segmentation.
      PubDate: THU, 25 JAN 2024 09:16:31 -04
      Issue No: Vol. 17, No. null (2024)
       
  • GEO SA-BSAR Synchronization and MTI Algorithm Based on Direct Signal and
           Clutter Subspace

    • Free pre-print version: Loading...

      Authors: Chang Cui;Xichao Dong;Zhiyang Chen;
      Pages: 4368 - 4378
      Abstract: Geosynchronous (GEO) spaceborne-airborne bistatic synthetic aperture radar, consisting of a GEO transmitter and an airborne multichannel receiver, is a potential moving target indication (MTI) system. However, such systems encounter synchronization challenges due to noncooperative transmission and reception. Besides, the inaccurate GEO orbital position for the MTI processor construction results in a low output signal-to-noise ratio (SNR) and reduced target detection probability. To address this, this article proposes a synchronization and MTI method based on direct signal and clutter subspace, which can enhance the output SNR for target detection even if the GEO orbital position is inaccurate. This method utilizes the direct signal to compensate for time and frequency synchronization errors. In addition, the residual error caused by the imprecise GEO orbit is estimated by approximating the clutter subspace. Next, a modified MTI processor is employed to suppress clutter and focus moving targets by using the residual error. Finally, the effectiveness of the proposed method is verified by numerical simulation experiments based on real IGSO orbital parameters.
      PubDate: THU, 25 JAN 2024 09:16:31 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Multispectral Satellite Image Generation Using StyleGAN3

    • Free pre-print version: Loading...

      Authors: Michael Alibani;Nicola Acito;Giovanni Corsini;
      Pages: 4379 - 4391
      Abstract: Satellite-based remote sensing images are essential for Earth surface analysis, serving diverse purposes in both civilian and military domains. Satellite images are used for analysis and decision making and are considered a reliable source of information. Recently, the field of image generation has developed increasingly sophisticated techniques, such as generative neural models, usually known as generative adversarial networks (GANs), to create synthetic images from scratch that appear almost real. Generative models have traditionally been applied to RGB or grayscale images and have been used for generating fake images of faces, animals, or objects. Currently, there are few studies regarding the application of GAN to multispectral satellite images. This work aims to test GAN models against the generation of multispectral satellite images, and in particular, the work explores the ability of the state-of-the-art StyleGAN3 model to produce high-quality synthetic Sentinel-2 images. The work delves into the configuration, training process, and evaluation of StyleGAN3 using custom Sentinel-2 datasets. StyleGAN3 results are compared with those provided by the proGAN model, the only GAN model tested so far on multispectral satellite data. Evaluation methods include visual inspection, spectral signature analysis, and a modified Fréchet inception distance for quantitative assessment. Results show that StyleGAN3 outperforms proGAN model exhibiting visually pleasing images. The quantitative comparison shows that StyleGAN3 provides the best results in terms of FID scores, in particular the improvement compared to proGAN increases as the spatial extent and spectral dimension of the generated images increases.
      PubDate: MON, 22 JAN 2024 09:21:31 -04
      Issue No: Vol. 17, No. null (2024)
       
  • An Efficient and Adaptive Reconstructive Homogeneous Block-Based Local
           Tensor Robust PCA for Feature Extraction of Hyperspectral Images

    • Free pre-print version: Loading...

      Authors: Longxia Qian;Xianyue Wang;Mei Hong;Hongrui Wang;Yongchui Zhang;
      Pages: 4392 - 4407
      Abstract: Model-driven tensor robust principal component analysis (TRPCA) has been widely applied to feature extraction of hyperspectral images (HSIs) and successfully protected two-dimensional spectral contextual information. Nevertheless, the current TRPCA-based feature extraction methods still destroy the underlying spectral and spatial–spectral joint contextual features. Moreover, these global iterative algorithms commonly ignore the heterogeneity of different real-world regions, increase the calculation burden, and improve practice operating time. To solve these issues, an efficient reconstructive homogeneous block-based local TRPCA is proposed for low-rank feature extraction, composed of a homogeneous block rebuilder and a local TRPCA low-rank feature extractor. The proposed local TRPCA is a novel data-model-driven algorithm depending on the data regulation. It remains the primary spatial and spectral contextual information and extracts the underlying homogeneity and heterogeneity characteristics of spatial, spectral, and spatial–spectral joint variables, which provides more essential features for further research than other model-driven TRPCA models. Furthermore, our local TRPCA feature extractor is an elaborate divide-and-rule model that executes on each homogeneous data block to extract low-rank features adaptively, remarkably decreasing computing cost and time. Experimental results on six hyperspectral datasets demonstrate that the proposed local TRPCA is more adaptive to HSIs and outperforms other state-of-the-art TRPCA-based feature extraction algorithms.
      PubDate: FRI, 19 JAN 2024 09:16:55 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Novel Category Discovery Without Forgetting for Automatic Target
           Recognition

    • Free pre-print version: Loading...

      Authors: Heqing Huang;Fei Gao;Jinping Sun;Jun Wang;Amir Hussain;Huiyu Zhou;
      Pages: 4408 - 4420
      Abstract: In this article, we explore a cutting-edge concept known as class incremental learning (CIL) in novel category discovery for synthetic aperture radar (SAR) targets (CNTs). This innovative task involves the challenge of identifying categories within unlabeled datasets by utilizing a provided labeled dataset as reference. In contrast to the conventional category discover approaches, our method introduces novel categories without relying on old labeled classes and effectively mitigates the issue of catastrophic forgetting. Specifically, to reduce the bias of the established categories toward unknown ones, CNT extracts representational information via self-supervised learning, gleaned directly from the SAR data itself to facilitate generalization. To retain the model's competence in classifying previously acquired knowledge, we employ a dual strategy incorporating the rehearsal of base category feature prototypes and the application of knowledge distillation. Our methodology integrates multiview and pseudolabeling strategies. In addition, we introduce a novel approach that focuses on enhancing the discernibility of class spaces. This strategy primarily ensures distinct separation of the unlabeled classes from base class prototypes, and imposes stringent constraints on the internal relationships among individual samples and their corresponding perspectives. To the best of our knowledge, this is the first study on category discovery in the CIL scenario. The experimental results show that our method significantly improves the performance on SAR images compared to the previous optimal method, which indicates the effectiveness of our method.
      PubDate: THU, 25 JAN 2024 09:16:31 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Hybrid Attention Fusion Embedded in Transformer for Remote Sensing Image
           Semantic Segmentation

    • Free pre-print version: Loading...

      Authors: Yan Chen;Quan Dong;Xiaofeng Wang;Qianchuan Zhang;Menglei Kang;Wenxiang Jiang;Mengyuan Wang;Lixiang Xu;Chen Zhang;
      Pages: 4421 - 4435
      Abstract: In the context of fast progress in deep learning, convolutional neural networks have been extensively applied to the semantic segmentation of remote sensing images and have achieved significant progress. However, certain limitations exist in capturing global contextual information due to the characteristics of convolutional local properties. Recently, Transformer has become a focus of research in computer vision and has shown great potential in extracting global contextual information, further promoting the development of semantic segmentation tasks. In this article, we use ResNet50 as an encoder, embed the hybrid attention mechanism into Transformer, and propose a Transformer-based decoder. The Channel-Spatial Transformer Block further aggregates features by integrating the local feature maps extracted by the encoder with their associated global dependencies. At the same time, an adaptive approach is employed to reweight the interdependent channel maps to enhance the feature fusion. The global cross-fusion module combines the extracted complementary features to obtain more comprehensive semantic information. Extensive comparative experiments were conducted on the ISPRS Potsdam and Vaihingen datasets, where mIoU reached 78.06% and 76.37%, respectively. The outcomes of multiple ablation experiments also validate the effectiveness of the proposed method.
      PubDate: MON, 29 JAN 2024 09:16:57 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Transformer and Convolutional Hybrid Neural Network for Seismic Impedance
           Inversion

    • Free pre-print version: Loading...

      Authors: Chunyu Ning;Bangyu Wu;Baohai Wu;
      Pages: 4436 - 4449
      Abstract: The inversion of elastic parameters especially P-wave impedance is an essential task in seismic exploration. Over the years, deep learning methods have made significant achievements in seismic impedance inversion, and convolutional neural networks (CNNs) become the dominating framework relying on extracting local features effectively. In fact, the elastic parameters temporal correlation consists of local and global characteristics, with the latter as a general trend in vertical direction due to gravity and diagenesis (vertical mechanical compression). Therefore, considering the excellent performance in capturing global dependencies of Transformer, we design an improved transformer encoder, a transformer and convolutional hybrid neural network (trans-CNN), for seismic impedance inversion. The designed network not only has the ability of transformer capturing global features with the facilitation of parallel computing but also the advantage of extracting local features of CNNs. With sparse well log data as labels, it can infer the absolute impedance from seismic data without an initial model. We also devise a relative time interval prediction self-supervised task to assist the network in better extracting seismic data features without adding any labels. Therefore, a multitask framework composed of self-supervised and supervised learning is used to train the network. We first conduct experiments on the Marmousi2 and overthrust model. The prediction profiles show that the proposed trans-CNN has better inversion and transfer learning ability than several comparable networks. We then test the proposed network on a field data, the experiments further suggest that trans-CNN can obtain stable inversion results with better horizontal continuity and high vertical resolution.
      PubDate: THU, 25 JAN 2024 09:16:31 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Edge-Enhanced GCIFFNet: A Multiclass Semantic Segmentation Network Based
           on Edge Enhancement and Multiscale Attention Mechanism

    • Free pre-print version: Loading...

      Authors: Long Chen;Zhiyuan Qu;Yao Zhang;Jingyang Liu;Ruwen Wang;Dezheng Zhang;
      Pages: 4450 - 4465
      Abstract: In recent years, remote sensing images (RSIs) have witnessed significant improvements in both quality and quantity. With the application of deep-learning techniques, these RSIs can be more effectively utilized to harnessed to aid in environment monitoring and urban planning. Semantic segmentation, as a common task in RSIs processing, confronts numerous challenges, including inaccurate classification, fuzzy boundaries, and other problems. This article proposes a novel semantic segmentation network known as the edge-enhanced global contextual information guided feature fusion network to address these challenges. This network consists of an edge-enhanced part and a backbone network part. First, in the encoding stage, the recurrent criss-cross attention block is employed, which incorporates spatial attention, mechanisms to capture global information. Second, in the decoding stage, a channel attention residual block module is proposed to facilitate the fusion of high-level and low-level features. Moreover, we enhance the network's ability to extract edge information during training by sharing parameters between the backbone and employing a specialized loss function. The network proposed in this article utilizes both channel attention and spatial attention at different stages, effectively utilizing edge information. Finally, we conduct experiments using the Yinchuan dataset and the LoveDA dataset. The experimental results show that the proposed network demonstrates excellent performance on both datasets.
      PubDate: TUE, 23 JAN 2024 09:18:08 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Multishot Compressive Hyperspectral Imaging Based on Tensor Fibered Rank
           Minimization and Its Primal-Dual Algorithm

    • Free pre-print version: Loading...

      Authors: Ting Xie;Xudong Kang;Renwei Dian;Tonghan Wang;Licheng Liu;
      Pages: 4466 - 4477
      Abstract: Coded aperture snapshot spectral imaging (CASSI) compresses tens to hundreds of spectral bands of the hyperspectral image (HSI) to a 2-D compressive measurement. For spatially or spectrally rich scenes, the compressive measurement provided by a single snapshot CASSI may not be sufficient. By taking multiple snapshots of the same scene, multishot CASSI leads to a less ill-posed inverse reconstruction problem, making the CASSI system more suitable for spatially or spectrally rich HSI. Considering the strong spectral correlation of HSI and the directional characteristics of mask shifting in multishot CASSI, the mode-1 tensor fibered rank (TFR) minimization is presented for its reconstruction in this article. Specifically, the mode-1 TFR is derived from the tensor singular value decomposition (t-SVD) to the mode-1 t-SVD, and the mode-1 TFR minimization is reduced to a mode-1 tensor nuclear norm minimization problem, to achieve more accurate HSI characterization in multishot CASSI reconstruction. The primal-dual algorithm (PDA) is applied to solve the objective optimization problem, which is flexible. Experimental results on the CAVE, Cuperite, and Urban datasets demonstrate the effectiveness of the proposed method.
      PubDate: MON, 29 JAN 2024 09:16:57 -04
      Issue No: Vol. 17, No. null (2024)
       
  • $C^{2}N^{2}$: Complex-Valued Contourlet Neural Network

    • Free pre-print version: Loading...

      Authors: Mengkun Liu;Licheng Jiao;Xu Liu;Lingling Li;Fang Liu;Shuyuan Yang;Yuwei Guo;Puhua Chen;
      Pages: 4478 - 4491
      Abstract: Complex-valued convolutional neural networks (CV-CNN) have recently gained recognition in feature representation learning. It implements the repeated application of the operations in convolution, local average pooling, and the absolute value of the resulting vectors. However, it is only conducted in the complex spatial domain, and lacks effective representation of directionality, singularity, and regularity in the complex spectral domain for anomaly detection of images. This is the key to feature learning representation of high-order singularity. To solve this problem, a complex-valued contourlet neural network (C$^{2}$N$^{2}$) is proposed in this article. It is novel in this sense that, different from the CV-CNN in the spatial domain, the spectral stream of C$^{2}$N$^{2}$ can enhance the multiresolution sparse representation of nonsubsampled contourlet (NSCT) with multiscales and multidirections for images. Furthermore, the spectral feature integration module is proposed to capture the statistical properties of the NSCT coefficients. It is shown that the proposed network can improve the distinguishability of feature learning and classification ability in theoretical analysis and experiments on three benchmark datasets (Flevoland, Xi'an, and Germany) compared with developed methods. Polarimetric synthetic aperture radar image classification is widely used in the fields of agriculture, forestry, and military. It must be emphasized that there is potential in effective feature learning representation and the generalization capability of C$^{2}$N$^{2}$ in deep learning, recognition, and interpretation.
      PubDate: FRI, 26 JAN 2024 09:16:24 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Spatio-Temporal Characteristics and Driving Mechanism of Alpine Peatland
           InSAR Surface Deformation—A Case Study of Maduo County, China

    • Free pre-print version: Loading...

      Authors: Yaoxiang Liu;Yi He;Wang Yang;Lifeng Zhang;
      Pages: 4492 - 4514
      Abstract: Surface deformation of alpine peatland in China has an important effect on runoff and is of great significance for wetland ecosystem protection. However, spatio-temporal characteristics of alpine peatland surface deformation in China lack systematic studies, and the driving mechanism is not yet clear. In this study, we selected the alpine peatland of Maduo County in China as the research object, surface deformation of peatland based on the small baseline subset radar interferometry technique was obtained, we analyzed spatio-temporal deformation characteristics and patterns of peatland, explored the driving mechanism of the peatland surface deformation with single-factor and multifactor combinations of Geo-detector, respectively. The results showed that the overall subsidence rates of peatlands in Maduo County, China slowed down year by year from 2018 to 2020, but there was seasonal freezing and thawing, subsidence rates of peatlands at high elevation and high slopes were stable, peatlands at low elevation and low slope were vulnerable to disturbance, subsidence rates are largest. Maliecuo, Bailongqu, and Gaerlawangzang regions were serious subsidence, the maximum subsidence rate was 159 mm/year. Meteorological factors and geological conditions were the main reasons for the surface deformation of alpine peatland in Maduo County, China. This study provides a theoretical basis for the conservation and restoration of peatland ecosystems in the alpine regions of China.
      PubDate: MON, 29 JAN 2024 09:16:57 -04
      Issue No: Vol. 17, No. null (2024)
       
  • GB-ArcSAR Imaging Based on Optimized Construction of Reference Plane

    • Free pre-print version: Loading...

      Authors: Yunkai Deng;Hanpu Zhou;Weiming Tian;Xin Xie;Cheng Hu;
      Pages: 4515 - 4522
      Abstract: Due to the specific arc-scanning way of ground-based arc-scanning synthetic aperture radar (GB-ArcSAR) to synthesize aperture, when a target's height is large enough and deviates much from the reference imaging plane, the image defocus would occur since that the target's range migration trajectory cannot be completely compensated. This article makes a theoretical analysis of the reasons for GB-ArcSAR imaging defocus, when it is utilized to observe those targets with large height differences, i.e., large elevation angles, from the imaging plane. A quantitative relationship between the imaging defocus level and the elevation angle is deduced with the simulation of an ideal target. Aiming at solving the problem of imaging defocus, an optimized imaging method of GB-ArcSAR based on the reference plane construction is proposed. A series of reference planes with different inclination angles or starting ranges are constructed and utilized for radar imaging separately. The minimum entropy criterion is then taken to search for the optimal plane. The qualitative and quantitative comparisons of the simulated and experimental datasets with other methods both prove the effectiveness of the proposed method.
      PubDate: FRI, 26 JAN 2024 09:16:24 -04
      Issue No: Vol. 17, No. null (2024)
       
  • An Effective Space-Borne ISAR High-Resolution Imaging Approach for
           Satellite On-Orbit Based on Minimum Entropy Optimization

    • Free pre-print version: Loading...

      Authors: Yifei Liu;Weidong Yu;Shenghui Yang;Shiqiang Li;
      Pages: 4523 - 4537
      Abstract: The space situational awareness program places great significance on obtaining high-resolution images of satellite in space orbit. By utilizing space-borne inverse synthetic aperture radar (SBISAR) can achieve high-resolution imaging of observed satellite (OS) on-orbit especially in geosynchronous orbit (GEO). However, the complex and nonuniform relative motion of the OS on-orbit and SBISAR produces 2-D spatial variant phase errors, which has a high-order form during long coherent processing intervals. Up to now, the imaging problem of SBISAR has not been effectively tackled. In this work, a novel method to compensate for the 2-D spatial variant phase errors for SBISAR imaging based on minimum entropy and quasi-Newton's method is proposed. First, the geometric model that considers the relative motion state of SBISAR and the OS on-orbit is established and the optimal observation time period is determined. Second, we propose the echo signal model for the satellite on-orbit and deduce the specific form of the high-order spatial variant phase errors. Third, based on image entropy, quasi-Newton's method is adopted to obtain the optimal solution for the phase error coefficients. Finally, a new initial estimation method that aims to overcome the local convergence of quasi-Newton's method is proposed. By utilizing the optimal parameters, the well-focused SBISAR image can be achieved. Taking GEO satellite imaging as an example, experiments based on scattering point simulation data verify the effectiveness of the proposed method.
      PubDate: MON, 29 JAN 2024 09:16:57 -04
      Issue No: Vol. 17, No. null (2024)
       
  • CD-CTFM: A Lightweight CNN-Transformer Network for Remote Sensing Cloud
           Detection Fusing Multiscale Features

    • Free pre-print version: Loading...

      Authors: Wenxuan Ge;Xubing Yang;Rui Jiang;Wei Shao;Li Zhang;
      Pages: 4538 - 4551
      Abstract: Clouds in remote sensing images inevitably affect information extraction, which hinders the following analysis of satellite images. Hence, cloud detection is a necessary preprocessing procedure. However, most existing methods have numerous calculations and parameters. In this article, a lightweight convolutional neural network (CNN)-Transformer network, CD-CTFM, is proposed to solve the problem, which is based on encoder–decoder architecture and incorporates the attention mechanism. In the encoder part, we utilize a lightweight network combing CNN and Transformer as backbone, which is conducive to extracting local and global features simultaneously. The backbone of CD-CTFM also incorporates attention gate based on dark channel extraction module. Moreover, a lightweight feature pyramid module is designed to fuse multiscale features with contextual information. In the decoder part, a lightweight channel-spatial attention module is integrated into each skip connection between encoder and decoder to extract low-level features while suppressing irrelevant information without introducing many parameters. Finally, the proposed model is evaluated on two cloud datasets, 38-Cloud and MODIS. The results demonstrate that CD-CTFM achieves comparable accuracy as the state-of-art methods and outperforms in terms of efficiency.
      PubDate: MON, 05 FEB 2024 09:16:30 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Quantifying Uncertainty in Slum Detection: Advancing Transfer Learning
           With Limited Data in Noisy Urban Environments

    • Free pre-print version: Loading...

      Authors: Thomas Stark;Michael Wurm;Xiao Xiang Zhu;Hannes Taubenböck;
      Pages: 4552 - 4565
      Abstract: In the intricate landscape of mapping urban slum dynamics, the significance of robust and efficient techniques is often underestimated and remains absent in many studies. This not only hampers the comprehensiveness of research but also undermines potential solutions that could be pivotal for addressing the complex challenges faced by these settlements. With this ethos in mind, we prioritize efficient methods to detect the complex urban morphologies of slum settlements. Leveraging transfer learning with minimal samples and estimating the probability of predictions for slum settlements, we uncover previously obscured patterns in urban structures. By using Monte Carlo dropout, we not only enhance classification performance in noisy datasets and ambiguous feature spaces but also gauge the uncertainty of our predictions. This offers deeper insights into the model's confidence in distinguishing slums, especially in scenarios where slums share characteristics with formal areas. Despite the inherent complexities, our custom CNN STnet stands out, delivering performance on par with renowned models like ResNet50 and Xception but with notably superior efficiency—faster training and inference, particularly with limited training samples. Combining Monte Carlo dropout, class-weighted loss function, and class-balanced transfer learning, we offer an efficient method to tackle the challenging task of classifying intricate urban patterns amidst noisy datasets. Our approach not only enhances artificial intelligence model training in noisy datasets but also advances our comprehension of slum dynamics, especially as these uncertainties shed light on the intricate intraurban variabilities of slum settlements.
      PubDate: MON, 29 JAN 2024 09:16:57 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Exploring the Linkages Between Different Types of Drought and Their
           Impacts on Crop Production in Kyrgyzstan

    • Free pre-print version: Loading...

      Authors: Sheng Chang;Erkin Isaev;Hong Chen;Bingfang Wu;Nana Yan;Zonghan Ma;Jihua Meng;
      Pages: 4566 - 4580
      Abstract: Drought is a perilous agrometeorological phenomenon that often causes crop damage in arid and semiarid regions vulnerable to climate variability. However, accurate drought monitoring remains deficient in many countries, including Kyrgyzstan, and the interconnections between several types of drought and contributions to crop yield are still unclear. Hence, we aimed to determine the propagation time in three types of drought (meteorological drought, soil drought, and vegetation drought) for understanding interconnections of them. Moreover, we focused on comprehensively evaluation the performance of multiple drought indices for each type over the complex terrain of Kyrgyzstan, especially for drought index of synergistic land surface temperature and vegetation conditions information. The results demonstrated that standard precipitation index (SPI) effectively detected meteorological drought, while the vegetation health index (VHI) coupled with temperature data was optimal for vegetation drought monitoring in Kyrgyzstan. Furthermore, our findings indicated a 1-month response time for soil drought at a 10 cm depth to SPI, and a 4-month response time at a 40 cm depth to meteorological drought (SPI). The response time of VHI to soil drought condition index (SMCI) was approximately 1 month, regardless of whether the soil drought occurred at a depth of 10 or 40 cm. In general, the response time of VHI to SPI was 3 months. Finally, by analyzing the correlation between crop yield productivity and drought indices, we discovered that the crop yield predictions by the three types of drought were differential and complex, but VHI was the most effective index. At the same time, VHIacc(May–Sep.), SMCIr(0–40 cm)_May–Sep., and SPI5_Aug. have different contributions to crop yield variations, and these are also differences in their impacts on different crops and provinces. The synergistic effect of the three types of drought may significantly improve crop yield prediction in Kyrgyzstan in future studies. These findings may significantly contribute to drought prevention and mitigation in drought-prone Central Asian countries.
      PubDate: MON, 29 JAN 2024 09:16:57 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Parameter Extraction Method of Overlapping Radar Signals Using Modulation
           Recognition-Guided Semantic Segmentation

    • Free pre-print version: Loading...

      Authors: Weibo Huo;Yang Luo;Hao Wang;Jifang Pei;Yin Zhang;Yulin Huang;Jianyu Yang;
      Pages: 4581 - 4596
      Abstract: Parameter extraction of radar signals is an important but challenging task in electronic warfare. In the modern electromagnetic environment, the radiation sources greatly increase, causing different radar signals to overlap, making the parameter extraction of radar signals difficult. Meanwhile, using radar signal parameter extraction methods that are not suitable for dealing with overlapping signals can lead to serious errors in this case. To address this, we propose a parameter extraction network for overlapping radar signals using modulation recognition-guided semantic segmentation. Specifically, we first design an encoder–decoder to segment overlapping radar signals, which uses channel rearrangement and modulation type filtering to increase the accuracy of segmentation. In this encoder–decoder, channel rearrangement is an optimization of convolution operation, aiming to increase the perceptual field while reducing feature information loss. And modulation type filtering can convert the results of semantic segmentation into masks corresponding to each radar signal, increasing the accuracy of segmentation. After the encoder–decoder, signal segmentation masks are obtained. Then, we compress these segmentation masks in the time and frequency domains, and extract the span of them to achieve accurate extraction of the pulsewidth and bandwidth of each radar signal. The experiments validate the feasibility of the proposed method.
      PubDate: MON, 05 FEB 2024 09:16:30 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Multiview Hypergraph Fusion Network for Change Detection in
           High-Resolution Remote Sensing Images

    • Free pre-print version: Loading...

      Authors: Xue Zhao;Kai Zhang;Feng Zhang;Jiande Sun;Wenbo Wan;Huaxiang Zhang;
      Pages: 4597 - 4610
      Abstract: Currently, convolutional neural networks and transformers have been the dominant paradigms for change detection (CD) thanks to their powerful local and global feature extraction capabilities. However, with the improvement of resolution, spatial, spectral, and temporal relationships among objects in remote sensing images are becoming more complicated and cannot be modeled efficiently by the existing methods. To capture the high-order complex relationships in images, we propose a multiview hypergraph fusion network (MVHFNet) for CD, in which the high-order relationships along spatial, spectral, and temporal views are extracted by hypergraph learning. Specifically, this network is composed of three branches, including the spectral hypergraph learning branch, the spatial hypergraph learning branch, and the temporal hypergraph learning branch. In these branches, multiview features are extracted by different attention modules, and hypergraph learning consisting of hypergraph construction and hypergraph convolution is imposed on these features to model the high-order relationships. Then, to integrate the multiview features from different branches, a multiview feature fusion module is designed, in which the multiview features are fused and condensed for the following prediction. Finally, the change map is produced by a prediction head. We conduct extensive experiments on three datasets, such as LEVIR-CD, SYSU-CD, and CLCD. The experimental results demonstrate that the proposed MVHFNet achieves better CD performance compared to some state-of-the-art methods.
      PubDate: WED, 31 JAN 2024 09:28:11 -04
      Issue No: Vol. 17, No. null (2024)
       
  • MGSFA-Net: Multiscale Global Scattering Feature Association Network for
           SAR Ship Target Recognition

    • Free pre-print version: Loading...

      Authors: Xianghui Zhang;Sijia Feng;Chenxi Zhao;Zhongzhen Sun;Siqian Zhang;Kefeng Ji;
      Pages: 4611 - 4625
      Abstract: Deep learning has offered new ideas in SAR ship target recognition. Although many methods improve the recognition performance through the improvement of loss function and migration of deep networks, scattering features as the important intrinsic features of SAR targets, need to be considered in the SAR ship recognition tasks. To introduce the scattering features into the deep network and characterize the features of ship targets more comprehensively, a multiscale global scattering feature association network (MGSFA-Net) for SAR ship target recognition is proposed in this article. In the network, the SAR ship target is first separated from the background by fine target segmentation. Then, the scattering centers (SCs) of ship targets are extracted and converted to local graph structures based on the $k$-nearest neighbors algorithm. The local graph structures are associated by the scattering center feature association module and enhanced by the multiscale feature enhancement module to produce the multiscale global scattering features. Moreover, the deep features of the targets are extracted by the multikernel deep feature extraction module to characterize the high-dimensional information. Finally, the scattering features and deep features are fused by weighted integration to enrich the diversity of features. The experimental results on the FUSAR-Ship and OpenSARShip dataset show that the MGSFA-Net can significantly improve the recognition performance, even on a few-shot condition with the accuracy increasing over 2%–3%. The feature distribution and visualization show the effectiveness of the MGSFA-Net to characterize the multiscale global scattering association features.
      PubDate: TUE, 23 JAN 2024 09:18:08 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Bidirectional Recurrent Imputation and Abundance Estimation of LULC
           Classes With MODIS Multispectral Time-Series and Geo-Topographic and
           Climatic Data

    • Free pre-print version: Loading...

      Authors: José Rodríguez-Ortega;Rohaifa Khaldi;Domingo Alcaraz-Segura;Siham Tabik;
      Pages: 4626 - 4645
      Abstract: Remotely sensed data are dominated by mixed land use and land cover (LULC) types. Spectral unmixing (SU) is a key technique that disentangles mixed pixels into constituent LULC types and their abundance fractions. While existing studies on deep learning (DL) for SU typically focus on single time-step hyperspectral or multispectral data, our work pioneers SU using MODIS MS time series, addressing missing data with end-to-end DL models. Our approach enhances a long-short-term-memory-based model by incorporating geographic, topographic (geo-topographic), and climatic ancillary information. Notably, our method eliminates the need for explicit endmember extraction, instead learning the input–output relationship between mixed spectra and LULC abundances through supervised learning. Experimental results demonstrate that integrating spectral-temporal input data with geo-topographic and climatic information significantly improves the estimation of LULC abundances in mixed pixels. To facilitate this study, we curated a novel labeled dataset for Andalusia (Spain) with monthly MODIS MS time series at 460-m resolution for 2013. Named Andalusia MultiSpectral MultiTemporal Unmixing, this dataset provides pixel-level annotations of LULC abundances along with ancillary information.
      PubDate: MON, 29 JAN 2024 09:16:57 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Mapping 30-m Resolution Bioclimatic Variables During 1991–2020 Climate
           Normals for Hubei Province, the Yangtze River Middle Reaches

    • Free pre-print version: Loading...

      Authors: Ruizhen Wang;Weitao Chen;Siyang Wan;Gaodian Zhou;Wenxi He;Lunche Wang;
      Pages: 4646 - 4662
      Abstract: High-resolution bioclimatic data are crucial to providing fine-scaled insights into biodiversity assessment, forestry, and agricultural management. Existing global bioclimatic datasets often exhibit kilometer-level coarse resolution or have miss the data in recent decades, potentially resulting in the issues of lower spatial accuracy, limited information, and restricted applicability in fine-scaled studies. Hubei Province in Yangtze River Middle Reaches has sparse meteorological stations in high-altitude mountainous areas to map the high-resolution bioclimatic variables directly. This study developed a 30-year averaged bioclimatic dataset for Hubei Province during 1991–2020 at a 30-m spatial resolution, utilizing monthly temperatures and precipitation derived from a downscaling-calibration framework. The downscaling of 1-km resolution climate variables was achieved by using a random forest model with 30-m resolution terrain and spatial covariates. Then, the geographical differential analysis was applied to improve the accuracy of downscaled products by including additional ground data. The mean absolute errors of calibrated monthly maximum, mean, minimum temperatures, and precipitation based on ordinary kriging decreased from 0.74 °C, 0.47 °C, 0.47 °C, and 28.27 mm to 0.43 °C, 0.28 °C, 0.36 °C, and 21.43 mm, respectively. Finally, calibrated climate variables were employed to calculate 19 annual bioclimatic variables, which were subsequently averaged over the 30-year period. The constructed bioclimatic dataset exhibits high overall consistency with the WorldClim dataset according to pixel-based comparison (Spearman correlation coefficients>0.6), with differences mainly attributed to the superior local accuracy of our dataset and climate changes. The dataset will provide fine-scaled, updated, and reliable data supports for local-related studies and decision making.
      PubDate: FRI, 26 JAN 2024 09:16:24 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Fast Array Ground Penetrating Radar Localization by CNN-Based Optimization
           Method

    • Free pre-print version: Loading...

      Authors: Changyu Zhou;Xu Bai;Li Yi;Munawar Shah;Motoyuki Sato;Xiaohua Tong;
      Pages: 4663 - 4673
      Abstract: This article presents an optimization-based approach to overcome redundancy arising from the multivariables enumeration process in multiple signal classification (MUSIC). By incorporating Broyden–Fletcher–Goldfarb–Shanno (BFGS) optimization, the computational speed of the MUSIC algorithm is significantly improved while maintaining mathematical accuracy. The optimization techniques require reasonable initial values to start the iteration, while for single target imaging purposes, the initial values can be acquired by the boundary between the near field and the far field. To generate suitable initial values for the optimization, we employ a modified convolutional neural network (CNN) to approximate the boundaries between the near and far fields, which vary with array system properties. Besides, the proposed method introduces a method for the Hessian matrix and gradient initialization for the BFGS method. Using simulation results as training samples, the modified CNN successfully establishes boundary approximations. Simulation and experimentation confirm the feasibility of our proposed method, showing its advantages in both accuracy and computation speed compared to existing techniques.
      PubDate: WED, 24 JAN 2024 09:17:27 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Mapping Water Clarity in Small Oligotrophic Lakes Using Sentinel-2 Imagery
           and Machine Learning Methods: A Case Study of Canandaigua Lake in Finger
           Lakes, New York

    • Free pre-print version: Loading...

      Authors: Rabia Munsaf Khan;Bahram Salehi;Milad Niroumand-Jadidi;Masoud Mahdianpari;
      Pages: 4674 - 4688
      Abstract: Optical remote sensing of water quality poses challenges in small oligotrophic lakes due to the diminishing contribution of constituents to the water-leaving radiance as water clarity increases. For monitoring water clarity over such lakes, this study utilizes machine learning models and data from citizen science to develop effective models for retrieving Secchi disk depth (SDD) in Canandaigua Lake, USA. Using Sentinel-2 band ratios as input, we trained random forest (RF), adaptive boosting, extreme gradient boosting, and support vector regression models using spatiotemporally distributed in situ data within 7 days of Senitnel-2 overpass. Each model was optimized using hyperparameter tuning, and cross-validation was used for accuracy assessment to compare the models’ effectiveness in retrieving SDD. The results indicate the superior performance of RF with an R2 of ∼0.74 and a root mean squared error of ∼0.72 m. A feature importance analysis for RF indicated the high relevance of the blue and green bands ratio in the estimation of SDD. The RF model was subsequently employed to generate temporal maps for Canandaigua Lake, indicating that water clarity tends to be higher during the early summer months (May and June) but declines during late summer and fall (September and October). This pattern can be closely associated with the increased algal presence in the lake. The spatial variability of the SDD indicated the possibility of greater sediments entering from the southern part of the lake. This study can be expanded to encompass other Finger Lakes, offering a comprehensive understanding of water clarity in these lake systems.
      PubDate: MON, 29 JAN 2024 09:16:57 -04
      Issue No: Vol. 17, No. null (2024)
       
  • CTMANet: A CNN-Transformer Hybrid Semantic Segmentation Network for
           Fine-Grained Airport Extraction in Complex SAR Scenes

    • Free pre-print version: Loading...

      Authors: Keyu Wu;Feng Cai;Haipeng Wang;
      Pages: 4689 - 4704
      Abstract: Airports represent essential infrastructure, offering substantial research and application potential. However, extracting airports from complex synthetic aperture radar (SAR) scenes is challenging due to the cluttered background and fine structure of airports. This necessitates the integration of global and local information for fine-grained extraction. To tackle this issue, this article introduces a novel framework for fine-grained extraction of airports from large-scale SAR images. First, a convolutional neural networks (CNN) transformer hybrid semantic segmentation network with multiscale contextual fusion is proposed, named CNN-transformer network (CTMANet). In this network, the encoder combines CNNs and transformers to capture local and global information, while the multiscale context aggregation block fuses multiscale contextual information. Skip connections between the encoder and decoder are established to minimize the loss of detailed information and fuse low-level features with high-level semantic features. Moreover, a category balance block is designed to address class imbalance. Experimental results on the GF-3 dataset demonstrate that CTMANet outperforms state-of-the-art methods, proving its superior suitability for fine-grained airport extraction in large-scale scenarios.
      PubDate: MON, 05 FEB 2024 09:16:30 -04
      Issue No: Vol. 17, No. null (2024)
       
  • SVSDet: A Fine-Grained Recognition Method for Ship Target Using Satellite
           Video

    • Free pre-print version: Loading...

      Authors: Shanwei Liu;Xi Bu;Mingming Xu;Hui Sheng;Zhe Zeng;Muhammad Yasir;
      Pages: 4726 - 4742
      Abstract: Target recognition from remote sensing images is commonly challenging because of large-scale variations and small objects, and these challenges are more prominent in satellite video images. The current object detection algorithms have some difficulties in fine-grained feature extraction and classification for multiscale and small objects. We propose a novel model called the SVSDet method based on YOLOv5 improvement to address the above-mentioned issues. In this method, we have introduced the space-to-depth module into the backbone of the network, which enhances the network's ability to extract fine-grained features. The neck structure is improved by using the bidirectional feature pyramid network to enhance the network's ability to extract features at multiple scales, thereby improving its overall multiscale feature extraction ability. Subsequently, we have replaced the C3 module in the original network's neck with the C2f module to obtain more abundant gradient flow information. This helps to improve the network's performance further. Finally, the coordinate attention module is introduced into the cross-scale feature connection path, which effectively enhances the network's target detection and recognition performance. We have conducted extensive comparative experiments and ablation experiments on the publicly available datasets ShipRSImageNet and SAT-MTB to confirm the effectiveness of our proposed SVSDet method. The performance of this approach is then evaluated using Jilin 1 satellite video data, and it outperforms the main YOLO series algorithms currently used.
      PubDate: MON, 29 JAN 2024 09:16:57 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Effects of Loss Function Choice on One-Shot HSI Target Detection With
           Paired Neural Networks

    • Free pre-print version: Loading...

      Authors: Kevin Benham;Phillip Lewis;Joseph C. Richardson;
      Pages: 4743 - 4750
      Abstract: Implementing reliable few-shot capable classifiers and detectors in machine learning is no trivial task and often requires parsing a large set of hyperparameters and training routine choices to find the best fit. One such choice is the loss function itself. In this effort, we study the validation and test performance of paired neural network (PNN) architectures using contrastive, hard-triplet, and semihard-triplet losses. These are tested by training multiple models to perform one-shot target detection on a custom synthetic hyperspectral image (HSI) dataset with and without reflectance calibration. We find that no single loss function is superior across all data treatments and standard scoring metrics can even disagree among the loss function choice among differing train, validation, and test split choices. We additionally analyze differences in detection map quality for selected test examples illustrating that while most are useful, some models will have more intuitive detection thresholds. Our work suggests multiple loss functions should be considered each time a new dataset and task are encountered to train PNNs for HSI target detection. These findings indicate significant variability in one-shot target detection performance based on the combination of training loss and data treatment but suggest the semihard-triplet loss, combined with a relatively simple reflectance calibration of the imagery, tends to generalize best across the common set of target materials studied.
      PubDate: THU, 25 JAN 2024 09:16:31 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Forest Disturbance Detection via Self-Supervised and Transfer Learning
           With Sentinel-1&2 Images

    • Free pre-print version: Loading...

      Authors: Rıdvan Salih Kuzu;Oleg Antropov;Matthieu Molinier;Corneliu Octavian Dumitru;Sudipan Saha;Xiao Xiang Zhu;
      Pages: 4751 - 4767
      Abstract: In this study, we examine the potential of leveraging self-supervised learning (SSL) and transfer learning methodologies for forest disturbance mapping using Earth Observation (EO) data. Our focus is on natural disturbances caused by windthrow and snowload damages. Particularly, we investigate the potential of knowledge-distillation-based contrastive learning approaches to obtain comprehensive representations of forest structure changes using Copernicus Sentinel-1 and Sentinel-2 satellite imagery. Leveraging pretrained backbone models from knowledge distillation, we employ transfer learning based on deep change vector analysis to delineate forest changes. We demonstrate developed approaches on two use cases, namely, mapping windthown forest using bitemporal Sentinel-1 and Sentinel-2 images, and mapping forest areas damaged by excessive snowload using bitemporal Sentinel-1 images. Developed self-supervised models were compared in a benchmarking exercise. The best results were provided by pixel-level contrastive learning for Sentinel-1-based snowload damage mapping with an overall accuracy of 84% and an $F_{1}$ score of 0.567, and for Sentinel-2-based forest windthrow mapping with an overall accuracy of 76.5% and an $F_{1}$ score of 0.692. We expect that developed methodologies can be useful for mapping also other types of forest disturbances using Copernicus Sentinel images or similar EO data. Our findings underscore the potential of SSL and transfer learning for enhancing forest disturbance detection using EO.
      PubDate: THU, 01 FEB 2024 09:16:33 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Interpretation and Attribution of Coastal Land Subsidence: An InSAR and
           Machine Learning Perspective

    • Free pre-print version: Loading...

      Authors: Xiaojun Qiao;Tianxing Chu;Evan Krell;Philippe Tissot;Seneca Holland;Mohamed Ahmed;Danielle Smilovsky;
      Pages: 4768 - 4783
      Abstract: Subsidence, the downward vertical land motion (VLM), plays a pivotal role in contributing to the risk of coastal flooding. Accurately estimating VLM and identifying its potential features related to subsidence can provide crucial information for stakeholders to make better-informed decisions. This study aimed to estimate large-scale subsidence at the Texas Gulf Coast and identify potential subsidence features using explainable models. Nine potential features were considered for modeling the VLM, ranging from natural terrain variations to anthropogenic activities. These features were used to train a random forest (RF) machine learning model. Explainable artificial intelligence (XAI) techniques including SHapley Additive exPlanations (SHAP) and impurity- and permutation-based feature importance were used to identify the contributions to subsidence. The results demonstrated favorable performance of the RF model, achieving an $R^{2}$ value of 0.56 during validation. XAI results underscored the significance of the digital elevation model in explaining subsidence at the Texas Coast. Additionally, XAI analysis highlighted the overall contribution of subsidence from anthropogenic activities, such as hydrocarbon extraction and groundwater withdrawal. Furthermore, the sample-level SHAP map provided detailed and reasonable subsidence-attribution results across the study area, showing potential for automatic and data-driven explanations of the VLM.
      PubDate: THU, 01 FEB 2024 09:16:34 -04
      Issue No: Vol. 17, No. null (2024)
       
  • FSOD4RSI: Few-Shot Object Detection for Remote Sensing Images via Features
           Aggregation and Scale Attention

    • Free pre-print version: Loading...

      Authors: Honghao Gao;Shuping Wu;Ye Wang;Jung Yoon Kim;Yueshen Xu;
      Pages: 4784 - 4796
      Abstract: Due to the continuous development of few-shot learning, there have been notable advancements in methods for few-shot object detection in recent years. However, most existing methods in this domain primarily focus on natural images, neglecting the challenges posed by variations in object scales, which are usually encountered in remote sensing images. This article proposes a new few-shot object detection model designed to handle the issue of object scale variation in remote sensing images. Our developed model has two essential parts: a feature aggregation module (FAM) and a scale-aware attention module (SAM). Considering the few-shot features of remote sensing images, we designed the FAM to improve the support and query features through channel multiplication operations utilizing a feature pyramid network and a transformer encoder. The created FAM better extracts the global features of remote sensing images and enhances the significant feature representation of few-shot remote sensing objects. In addition, we design the SAM to address the scale variation problems that frequently occur in remote sensing images. By employing multiscale convolutions, the SAM enables the acquisition of contextual features while adapting to objects of varying scales. Extensive experiments were conducted on benchmark datasets, including NWPU VHR-10 and DIOR datasets, and the results show that our model indeed addresses the challenges posed by object scale variation and improves the applicability of few-shot object detection in the remote sensing domain.
      PubDate: TUE, 06 FEB 2024 09:19:27 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Common Practices and Taxonomy in Deep Multiview Fusion for Remote Sensing
           Applications

    • Free pre-print version: Loading...

      Authors: Francisco Mena;Diego Arenas;Marlon Nuske;Andreas Dengel;
      Pages: 4797 - 4818
      Abstract: The advances in remote sensing technologies have boosted applications for Earth observation. These technologies provide multiple observations or views with different levels of information. They might contain static or temporary views with different levels of resolution, in addition to having different types and amounts of noise due to sensor calibration or deterioration. A great variety of deep learning models have been applied to fuse the information from these multiple views, known as deep multiview (MV) or multimodal fusion learning. However, the approaches in the literature vary greatly since different terminology is used to refer to similar concepts or different illustrations are given to similar techniques. This article gathers works on MV fusion for Earth observation by focusing on the common practices and approaches used in the literature. We summarize and structure insights from several different publications concentrating on unifying points and ideas. In this manuscript, we provide a harmonized terminology while at the same time mentioning the various alternative terms that are used in literature. The topics covered by the works reviewed focus on supervised learning with the use of neural network models. We hope this review, with a long list of recent references, can support future research and lead to a unified advance in the area.
      PubDate: FRI, 02 FEB 2024 09:16:25 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Modified Oriented Dihedral Model for Scattering Characteristic Description
           With PolSAR Data

    • Free pre-print version: Loading...

      Authors: Yifan Chen;Lamei Zhang;Jordi J. Mallorqui;Bin Zou;
      Pages: 4829 - 4844
      Abstract: Dihedral is a common structure in polarimetric SAR images and can be found on many man-made targets. Many researchers have proposed different dihedral models, but the accuracy of these models is limited. In this case, the feature extraction methods based on these models are also not effective enough, which affects the subsequent applications such as target detection. Therefore, it is necessary to propose a new and accurate scattering model, which can be applied to dihedral with different orientation angles for feature extraction and target detection. In this article, a general scattering model called modified oriented dihedral scattering model (MODM) is proposed based on physical optics and geometric optics of high-frequency approximation techniques. By analyzing the propagation and reflection of electromagnetic wave, MODM can accurately describe the scattering characteristic of dihedral for all observation conditions. In order to apply the model to real polarimetric synthetic aperture radar images, MODM is introduced into a new feature extraction method, which is called five-scattering component polarimetric decomposition method (MODM-5SD). Feature extraction and target detection experiments of buildings with various oriented dihedral structures are performed using different datasets, which show that dihedral scattering components from oriented dihedral structures can be more effectively extracted by MODM-5SD. In addition, more buildings with oriented dihedral structures can be detected with the features from MODM-5SD. The experimental results show that MODM can more accurately describe the scattering characteristic of dihedral, which can be further applied for scattering characterization and feature extraction of targets with typical dihedral structures.
      PubDate: FRI, 19 JAN 2024 09:16:55 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Optimized GNSS Cal/Val Site Selection for Expanding InSAR Viability in
           Areas With Low Phase Coherence: A Case Study for Southern Louisiana

    • Free pre-print version: Loading...

      Authors: Bhuvan K. Varugu;Cathleen E. Jones;Ke Wang;Jingyi Chen;Randy L. Osborne;George Z. Voyiadjis;
      Pages: 4875 - 4889
      Abstract: Interferometric synthetic aperture radar (InSAR) techniques can be used to derive spatially dense “relative” measurements of vertical land motion (VLM), whereas global navigation satellite system (GNSS) provides point-based “absolute” measurements of VLM. The combination of GNSS and InSAR observations can yield spatially dense VLM measurements in an absolute reference frame. In addition, GNSS observations can be used to correct atmospheric noise in InSAR deformation measurements and serve as a complementary measure to isolate deep and shallow subsidence components. Given the increasing spatial and temporal coverage available from InSAR satellites, there is a need to establish calibration/validation networks that enable the use of InSAR for measuring VLM in coherence-challenged areas such as many low-lying coastal lands. In this study, we provide a method for the selection of sites for new GNSS installations such that the resulting GNSS network can better serve as tie points and validation for InSAR in areas where low coherence prevents high-fidelity phase unwrapping. Our method is applied in a case study for expanding the existing GNSS network in southern Louisiana, using abandoned oil well sites as potential sites. Considering practical limitations, distribution among various land classes, and following National Geodetic Survey guidelines, our proposed GNSS network consists of 61 (45 existing + 16 new) stations spread over a 50 000 km2 area of southern Louisiana.
      PubDate: TUE, 06 FEB 2024 09:19:27 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Optimization of Random Surface Scattering Models for RR Polarization in
           SoOp-R/GNSS-R Applications

    • Free pre-print version: Loading...

      Authors: Xuerui Wu;Lixiong Chen;Jiancheng Shi;
      Pages: 4890 - 4898
      Abstract: Polarization in global navigation satellite system-reflectometry (GNSS-R) or signal of opportunity-reflectometry (SoOP-R) is commonly used for retrieving geophysical parameters. However, the attention toward other polarizations of reflected signals has increased with developments in this field. The widely used equation for RR polarization suggests that it decreases as soil moisture content increases, which contradicts the experimental data. The accurate forward calculation of RR polarization is essential for the subsequent retrieval algorithm in polarization GNSS-R/ SoOP-R. To address this issue, three new models have been developed: specular reflectivity model for polarization GNSS-R (Spec4PolR), small perturbation model for polarization GNSS-R (SPM4Pol), and Umich model for polarization GNSS-R (Umich4PolR). The Mueller matrix of these three models has been presented, and the wave synthesis technique has been employed to calculate the reflectivity at RR polarization. Spec4polR uses only three elements in the Mueller matrix for final reflectivity, while five elements are used in Umich4polR. In SPM4Pol, all elements construct the Mueller matrix, and only nine elements are employed for calculation. The effects of each element on soil moisture content are presented, and the final reflectivity at RR polarization is illustrated. However, due to the simple formulation of Spec4Pol, its reflectivity at RR polarization still decreases as soil moisture content increases. On the other hand, the results of SPM4Pol and Umich4Pol are consistent with the measured data, and the reflectivity at RR polarization increases as soil moisture content increases. The formula developed in this article for calculating RR polarization will contribute to subsequent polarization studies and geophysical parameter retrieval based on RR polarization.
      PubDate: THU, 15 FEB 2024 09:17:30 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Bitemporal Attention Transformer for Building Change Detection and
           Building Damage Assessment

    • Free pre-print version: Loading...

      Authors: Wen Lu;Lu Wei;Minh Nguyen;
      Pages: 4917 - 4935
      Abstract: Building change detection (BCD) holds significant value in the context of monitoring land use, whereas building damage assessment (BDA) plays a crucial role in expediting humanitarian rescue efforts post-disasters. To address these needs, we propose the bitemporal attention module (BAM) as an innovative cross-attention mechanism aimed at effectively capturing spatio-temporal semantic relations between a pair of bitemporal remote sensing images. Within BAM, a shifted windowing scheme has been implemented to confine the scope of the cross-attention mechanism to a specific range, not only excluding remote and irrelevant information but also contributing to computational efficiency. Moreover, existing methods for BDA often overlook the inherent order of ordinal labels, treating the BDA task simplistically as a multiclass semantic segmentation problem. Recognizing the vital significance of ordinal relationships, we approach the BDA task as an ordinal regression problem. To address this, we introduce a rank-consistent ordinal regression loss function to train our proposed change detection network, bitemporal attention transformer. Our method achieves state-of-the-art accuracy on two BCD datasets (LEVIR-CD+ and S2Looking), as well as the largest BDA dataset (xBD).
      PubDate: TUE, 16 JAN 2024 09:16:09 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Water-Body Type Classification in Dual PolSAR Imagery Using a Two-Step
           Deep-Learning Method

    • Free pre-print version: Loading...

      Authors: Qiming Yuan;Jing Lu;Lin Wu;Yabo Huang;Zhengwei Guo;Ning Li;
      Pages: 4966 - 4985
      Abstract: Water-body type problems classification plays a vital role in ecological conservation, water resource management, and urban planning. Accurate classification can aid decision-makers in understanding the functions of different water-body types, providing key information for urban planning and promoting harmony between human activities and the natural environment. Despite extensive research in the field of water-body segmentation, exploration in the water-body type classification community is not as widespread. Therefore, this article proposes a novel water-body type classification method based on a two-step deep-learning model, decomposing water-body type classification into water-body segmentation and water-body type identification. Especially, this method constructs a unique data strategy by organically integrating backscatter features, polarimetric features, and DEM features, providing the model with rich and comprehensive information. In the first step, the segmentation network uses the fused feature to extract all water-body from synthetic aperture radar images. Subsequently, the extracted water-body are combined with the input data, forming a multifeature input for the identification network to distinguish between natural and artificial water-body. During this process, a swarm intelligence optimization algorithm is employed to explore the optimal hyperparameters of the network, including those of the segmentation and identification networks. Finally, the proposed method is assessed using extensive experiments on water-body segmentation tasks, water-body type identification tasks, and joint water-body type classification tasks. This article not only provides a new perspective in the field of water-body type classification but also demonstrates the immense potential of deep-learning network hyperparameter optimization and feature fusion in solving such.
      PubDate: THU, 01 FEB 2024 09:16:33 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Earth Video Cube: A Geospatial Data Cube for Multisource Earth Observation
           Video Management and Analysis

    • Free pre-print version: Loading...

      Authors: Zilong Li;Zhipeng Cao;Peng Yue;Chenxiao Zhang;
      Pages: 4986 - 5000
      Abstract: Earth observation (EO) videos are undergoing rapid expansion due to the swift advancements in aerial, spaceborne, and ground remote sensing technologies enabling the continuous capture of imagery of the Earth's surface. Compared with traditional image-based EO data, EO videos offer persistent EO, rendering a promising observation resource across diverse applications, including climate monitoring and hazard assessment. The continuous observation capability introduces a challenge to the community, i.e., how to effectively manage and fully harness the value of the substantial volume of EO videos. In this article, we propose a novel approach leveraging a spatiotemporal data cube with EO video management to facilitate analysis. It suggests an analysis ready data (ARD) for EO videos, termed as analysis ready video data, which is incorporated into an Earth video cube. The ARD includes semantics at frame, object/trajectory, and event levels. This article presents the cube data organization and query processing for EO videos. A prototype system is implemented to demonstrate the applicability of the approach.
      PubDate: THU, 25 JAN 2024 09:16:31 -04
      Issue No: Vol. 17, No. null (2024)
       
  • Improved Forest Signal Detection for Space-Borne Photon-Counting LiDAR
           Using Automatic Machine Learning

    • Free pre-print version: Loading...

      Authors: Bo Zhang;Li Zhang;Yong Pang;Peter North;Min Yan;Hongge Ren;Linlin Ruan;Zhenyu Yang;Bowei Chen;
      Pages: 1 - 13
      Abstract: NASA's (National Aeronautics and Space Administration) ICESat-2 with a Photon Counting LiDAR (Light Detection And Ranging) Sensor sensitively detects signal photons at high speed with an advanced detection system called the Advanced Topographic Laser Altimeter System (ATLAS). However, the sensor also extracts a large amount of background photon noise coming from the atmosphere, ground, sun, or other radiation. This condition is particularly evident in forest areas. This study proposes an automatic machine learning approach to utilize data for forestry applications to improve data availability compared to NASA's official product. Our method uses only a very limited number (10%) of sample points for training, ensuring operational efficiency and training accuracy. We conclude that the integrated learning performance generally outperforms single models, and the mean F1 score of all tests is approximately 0.9. The mean F1 score of the Stacked Ensembles model is 0.957 ahead of the other models. The top three variables used in training models are kNNdist5, kNNdist10, and h. These three variables could explain 51.6% of the components of the models. Over the regions tested, the proposed method could improve the proportion of signals correctly identified by 6.4%, 12.2%, 2.7%, 9.3%, and 1.4% in five datasets. The model performs better in low signal-to-noise (SNR) datasets less than 7.5. The method would be largely unaffected by differences in topography, noise distribution, and SNR. The classifiers could correct misclassified labels in ATL08 products and show good stability in different conditions.
      PubDate: THU, 29 JUN 2023 10:01:47 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Deep Hyperspectral Shots: Deep Snap Smooth Wavelet Convolutional Neural
           Network Shots Ensemble for Hyperspectral Image Classification

    • Free pre-print version: Loading...

      Authors: Farhan Ullah;Yaqian Long;Irfan Ullah;Rehan Ullah Khan;Salabat Khan;Khalil Khan;Maqbool Khan;Giovanni Pau;
      Pages: 14 - 34
      Abstract: The deployment of convolutional neural networks (CNNs) to classify hyperspectral images is extensively discussed in the research study. A number of different algorithms and approaches are applied, including 2-D CNN, 3-D CNN, support vector machine (SVM), regression models, and other state-of-the-art deep learning models, although these methods do not show good performance for hyperspectral image classification. Furthermore, 3-D CNNs require a lot of computational power and are not mainly employed, whereas 2-D CNNs do not constitute multiresolution image processing and exclusively focus on spatial features. However, 3D–2D CNNs aim to incorporate spectral and spatial features, and their efficiency while being evaluated on various datasets tends to be limited. Moreover, a number of deep learning models have been proposed recently, but their performance is still limited. In order to solve these problems, in this article, we propose a novel deep hyperspectral shot, a deep smooth wavelet CNN shots ensemble for hyperspectral image classification. A deep smooth wavelet CNN utilizes layers of wavelet transform to extract spectral features. The computation of a wavelet transform is less intensive as compared to the computation of a 3-D CNN. After that, the extracted spectral features are integrated into 2-D CNN, which generates spatial features, as a result, generates a spatial–spectral feature vector for classification. Furthermore, we introduce the snapshots generation method and employ a cyclic annealing schedule to converge to several local minima along its optimization path and save the models. We build several snapshots of the deep hyperspectral shots model to enhance the performance of our proposed method. We propose the snapshots optimization and ensemble selection approach in order to solve the optimization problem within ensemble creation and further enhance the performance. In addition, we also introduce a novel activation function called Relish to increase spatial–spectral feature propagation and advance for smoother gradients. Overall, we ensemble the snapshots of our proposed method and achieved that can classify multiresolution HSI data with high accuracy. Experiments performed on benchmark datasets, our proposed method, deep hyperspectral shots, achieved overall accuracies of 99.96%, 97.91%, and 99.49% on the Salinas, Indian Pines, and Pavia University datasets against the state-of-the-art methods.
      PubDate: WED, 13 SEP 2023 10:01:35 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Bistatic Rough Surface Scattering at P-Band in Grand Mesa Based on Lidar
           Observations of Surface Roughness and Topography

    • Free pre-print version: Loading...

      Authors: Haokui Xu;Leung Tsang;Xiaolan Xu;Simon Yueh;Steven A. Margulis;Rashmi Shah;
      Pages: 35 - 44
      Abstract: In this article, we use the analytical Kirchhoff solution (AKS) and numerical Kirchhoff approach to study the bistatic scattering field ($\gamma $) from mountain terrain at P-band frequency. The study area is Grand Mesa, Colorado, USA, and the properties of land surface roughness are extracted from airborne lidar surveys. The bistatic scattering coefficient $\gamma $ of variance fields, denoted by${\gamma }_v$, for several cases of radar resolutions over a 3.6 km by 3.6 km area are calculated at various scattering azimuth angles. Based on the lidar measurements, the land surface is decomposed into ${f}_2 + {f}_3$, where ${f}_3$ is 30 m of deterministic planar patches to approximate the coarse topography and ${f}_2$ is modeled by random rough surfaces with correlation functions. Surface roughness statistics derived from the Lidar data give a typical root mean square height of 0.07 m and a correlation length of 3.6 m for ${f}_2$. The mean values of slopes of ${f}_3$ are 1.3° and 0° with a standard deviation of 1° each, respectively in the x and y directions. Simulations using AKS show that the values of bistatic scattering coefficients for the variance of scattered fields can reach above 10 dB over a range of azimuth angles ${\phi }_s$ in the vicinity of the specular direction. Even in mountainous regions, the value of the ${\gamma }_v$ around the forward scattering direction is much larger than that for radar backscattering, and thus could support the use of a synthetic aperture radar concept based on signals of opportunity with data acquisition near the forward direction.
      PubDate: THU, 12 OCT 2023 09:16:21 -04
      Issue No: Vol. 17, No. null (2023)
       
  • A Novel Transformer Network With Shifted Window Cross-Attention for
           Spatiotemporal Weather Forecasting

    • Free pre-print version: Loading...

      Authors: Alabi Bojesomo;Hasan AlMarzouqi;Panos Liatsis;
      Pages: 45 - 55
      Abstract: Earth observation is a growing research area that can capitalize on the powers of artificial intelligence for short time forecasting, a now-casting scenario. In this work, we tackle the challenge of weather forecasting using a video transformer network. Vision transformer architectures have been explored in various applications, with major constraints being the computational complexity of attention and the data-hungry training. To address these issues, we propose the use of video Swin-transformer (VST), coupled with a dedicated augmentation scheme. Moreover, we employ gradual spatial reduction on the encoder side and cross-attention on the decoder. The proposed approach is tested on the Weather4Cast2021 weather forecasting challenge data, which requires the prediction of 8 h ahead future frames (4 per hour) from an hourly weather product sequence. The dataset was normalized to 0–1 to facilitate the use of the evaluation metrics across different datasets. The model results in an mse score of 0.4750 when provided with training data, and 0.4420 during transfer learning without using training data, respectively.
      PubDate: FRI, 13 OCT 2023 09:16:31 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Cam-PC: A Novel Method for Camouflaging Point Clouds to Counter
           Adversarial Deception in Remote Sensing

    • Free pre-print version: Loading...

      Authors: Bo Wei;Teng Huang;Xi Zhang;Jiaming Liang;Yunhao Li;Cong Cao;Dan Li;Yongfeng Chen;Huagang Xiong;Feng Jiang;Xiqiu Zhang;
      Pages: 56 - 67
      Abstract: Synthetic aperture LiDAR can generate point cloud data, which is widely used in 3-D scene reconstruction. However, existing point cloud object recognition methods are vulnerable to adversarial attacks, and such attacks are difficult to transfer to the physical world. Even if adversarial perturbations are added to physical objects, they are easily detectable by other sensors. Our proposed method includes two modules, R-D and D-R, which generate more concealed adversarial point cloud samples by modifying digital and physical features. The R-D module maps real-world entities to point cloud data in the digital world and generates adversarial samples by modifying signal amplitude values. The D-R module constructs adversarial objects by modifying the surface diffuse reflectance of the target object based on ray tracing and correspondences between digital and physical features. Our method is evaluated through experiments on attack effectiveness, robustness after subsampling and transferability, demonstrating its effectiveness, and achieving new state-of-the-art performance.
      PubDate: FRI, 13 OCT 2023 09:16:31 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Subretrograde Geosynchronous SAR: Parameter Design and Performance
           Analysis

    • Free pre-print version: Loading...

      Authors: Caipin Li;Qingjun Zhang;Jiao Liu;Guoqiang Han;Bo Liu;Chongdi Duan;Zheng Lu;
      Pages: 68 - 83
      Abstract: In recent years, geosynchronous synthetic aperture radar (GEO SAR) has attracted many scientists to carry out relevant research. Its research is in-depth and gradually maturing. However, conventional GEO SAR has the problems of great influence by the Earth's rotation, complex timing design, uneven flight speed in the whole orbit, nonorthogonal range, and azimuth resolution, which limits its application. A new concept of subretrograde geosynchronous synthetic aperture radar is proposed in this article. Compared with traditional GEO SAR, the satellite has the advantages of less influence by the Earth's rotation, simple timing, uniform motion speed in the whole orbit, shorter revisit time, better resolution, and small space-variant in azimuth and range. The radar parameters of subretrograde GEO SAR are designed. Through the research, it is found that the system sensitivity and resolution no longer change with the orbit position, and its timing design does not need to consider the range migration variation. In addition, the imaging characteristics of subretrograde GEO SAR are investigated. The results show that the range model can meet the imaging requirements as long as it is expanded to the fourth order. After that, the influence of orbit perturbation error in subretrograde GEO SAR is given for the first time. Last, two ways to solve the observation blind area of subretrograde GEO SAR are proposed, one is to change the orbit inclination, and another way is to use the bistatic mode.
      PubDate: MON, 16 OCT 2023 09:17:35 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Coherent Change Detection for High-Resolution Drone-Borne SAR at 24 GHz

    • Free pre-print version: Loading...

      Authors: Ali Bekar;Muge Bekar;Christopher J. Baker;Michail Antoniou;
      Pages: 84 - 94
      Abstract: This article develops and examines techniques to enable coherent change detection (CCD) for short-range, high-resolution drone-borne synthetic aperture radar (SAR) systems operating at high frequencies. The potential of using high frequencies at short ranges for fine-resolution imagery and sensitivity to temporal change detection is highlighted, as are the challenges in terms of sensitivity to SAR motion errors. SAR system characteristics for CCD are derived, and the impact of motion errors, which leads to spatial decorrelation and co-registration errors on CCD maps, is discussed. Subsequently, a CCD algorithm able to generate change maps with a>0.75 average coherence value is presented. The validity of the approach is tested through various experimental scenarios. As a result, car tyre marks and human footprints are possible to discern with a drone-borne SAR demonstrator operating at 24 GHz.
      PubDate: TUE, 17 OCT 2023 09:17:24 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Baseline Optimization for Spatial-Temporal Coupling in Geosynchronous
           Differential Synthetic Aperture Radar Tomography

    • Free pre-print version: Loading...

      Authors: Xichao Dong;Yan Liu;Zhiyang Chen;Xinyan Chen;Yuanhao Li;Cheng Hu;Feifeng Liu;
      Pages: 95 - 108
      Abstract: Differential synthetic aperture radar tomography (D-TomoSAR) uses multiple SAR acquisitions at different times to form an elevation-time 2-D synthetic aperture, enabling recovery of the target's 3-D structure and deformation velocity. However, the imaging performance of D-TomoSAR is influenced by the spatial-temporal baseline distribution, which is one of the crucial factors. Due to the difference in orbit altitude, the effects of various perturbation forces on geosynchronous SAR (GEO SAR) are significantly different from those on low Earth orbit SAR (LEO SAR), leading to different geometries in the spatial-temporal baseline distributions of GEO SAR and LEO SAR in the repeat-pass acquisitions. The spatial-temporal baseline distribution of LEO SAR is random, while that of GEO SAR is coupled. Although GEO SAR obtains a large number of acquisitions due to the short repeat-pass time, using data under all spatial-temporal baselines for D-TomoSAR does not necessarily provide the best imaging results. Therefore, how to select spatial-temporal baselines that can improve the estimation accuracy is important. In this article, the Cramér–Rao lower bound of GEO D-TomoSAR estimation accuracy is determined by considering multiple factors, including baseline decorrelation and spatial-temporal coupling. The optimal spatial-temporal baseline selection is modeled as a multiobjective problem and solved by the Nondominated Sorting Genetic Algorithm II (NSGA-II). Given the absence of in-orbit GEO SAR and the orbital similarity between GEO SAR and BeiDou Inclined Geosynchronous Orbit (IGSO) satellite, the real ephemeris of BeiDou IGSO satellite is used to simulate the baseline distribution of GEO SAR and verify the baseline optimization method.
      PubDate: WED, 18 OCT 2023 09:16:59 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Effects of Atmospheric, Topographic, and BRDF Correction on Imaging
           Spectroscopy-Derived Data Products

    • Free pre-print version: Loading...

      Authors: Marius Vögtli;Daniel Schläpfer;Meredith C. Schuman;Michael E. Schaepman;Mathias Kneubühler;Alexander Damm;
      Pages: 109 - 126
      Abstract: Surface reflectance is an important data product in imaging spectroscopy for obtaining surface information. The complex retrieval of surface reflectance, however, critically relies on accurate knowledge of atmospheric absorption and scattering, and the compensation of these effects. Furthermore, illumination and observation geometry in combination with surface reflectance anisotropy determine dynamics in retrieved surface reflectance not related to surface absorption properties. To the best of authors' knowledge, no comprehensive assessment of the impact of atmospheric, topographic, and anisotropy effects on derived surface information is available so far. This study systematically evaluates the impact of these effects on reflectance, albedo, and vegetation products. Using three well-established processing schemes (ATCOR F., ATCOR R., and BREFCOR), high-resolution APEX imaging spectroscopy data, covering a large gradient of illumination and observation angles, are brought to several processing states, varyingly affected by mentioned effects. Pixel-wise differences of surface reflectance, albedo, and spectral indices of neighboring flight lines are quantitatively analyzed in their respective overlapping area. We found that compensation of atmospheric effects reveals actual anisotropy-related dynamics in surface reflectance and derived albedo, related to an increase in pixel-wise relative reflectance and albedo differences of more than 40%. Subsequent anisotropy compensation allows us to successfully reduce apparent relative reflectance and albedo differences by up to 20%. In contrast, spectral indices are less affected by atmospheric and anisotropy effects, showing relative differences of 3% to 10% in overlapping regions of flight lines. We recommend to base decisions on the use of appropriate processing schemes on individual use cases considering envisioned data products.
      PubDate: THU, 19 OCT 2023 09:16:42 -04
      Issue No: Vol. 17, No. null (2023)
       
  • A Novel Approach Extracting Extreme Points Combining Multidirection Phase
           

    • Free pre-print version: Loading...

      Authors: Haiqiao Liu;Zichao Gong;Qin Wan;Wei Huang;Jun Yu;Meng Liu;
      Pages: 127 - 137
      Abstract: This article proposed the novel approach of multimodel image matching based on the feature integration of multidirectional phase superposition and weighted moment diagrams to handle the matching problem of large contrast differences, lots of noises, and nonlinear radiometric distortion between an image pair. There are three steps for this method. First, the presented method extracted the local extreme points in multidirectional phase superposition diagram and the edge extreme points in the weighted moment diagram. Then, the characteristic descriptor was produced. Finally, angle-assisted difference measurement was put forward to handle the problem of angle reverse. In the collected dataset, this method was experimentally compared with RIFT, CoFSM, and HAPCG. The number of extreme points and Recall of PSMWD are higher than those of RIFT and HAPCG. In image pairs of noises, the number of correct matchings of PSMWD is about twice of HAPCG. Therefore, the proposed theory has practical importance.
      PubDate: THU, 19 OCT 2023 09:16:42 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Global Priors With Anchored-Stripe Attention and Multiscale Convolution
           for Remote Sensing Image Compression

    • Free pre-print version: Loading...

      Authors: Lei Zhang;Xugang Hu;Tianpeng Pan;Lili Zhang;
      Pages: 138 - 149
      Abstract: Compressing remote sensing images with high spatial and spectral resolution plays an important role in subsequent image processing and information acquisition. Accurate data modeling can help the entropy model to better estimate the entropy value. For better image recovery, it is necessary to make full use of the prior information contained in the latent information. To achieve global association and hierarchical modeling of latent elements, this article proposes adding additional global anchored-stripe self-attention capturing global, local, and interchannel dependencies. To enhance the feature extraction capabilities of the encoder and the decoder, the multiscale attention module of depthwise convolution is used to increase the receptive field and nonlinear conversion process, ensuring that the network can retain more useful information. We evaluate the compression performance of the proposed method in terms of rate–distortion curves and running speed. Through comparative experiments on DOTA, LoveDA, and UC-Merced datasets, it is shown that the proposed method has a faster running speed than that of the context model. It outperforms some traditional compression methods, such as BPG, WebP, JPEG2000, and state-of-the-art deep-learning-based methods, in terms of peak signal-to-noise ratio and multiscale structural similarity index measure. In terms of perceptual quality, adding perceptual loss reduces the smooth image blurring due to MSE loss, and the proposed method has better image perceptual quality under the approximate bits per pixel.
      PubDate: MON, 23 OCT 2023 09:20:57 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Change Detection Enhanced by Spatial-Temporal Association for Bare Soil
           Land Using Remote Sensing Images

    • Free pre-print version: Loading...

      Authors: Sasha Wu;Yalan Liu;Shufu Liu;Dacheng Wang;Linjun Yu;Yuhuan Ren;
      Pages: 150 - 161
      Abstract: As dust source bare soil land (BSL) contributes to air pollution and affects the photosynthesis of green plants and carbon absorption, it is the objective of this study to develop an approach for monitoring the changes of BSL using remote sensing technology. Unlike other land use/cover types, the classification of BSL as well as its change detection is often ignored. For traditional convolutional neural networks, deep layers cause a long range between input and output, inevitably leading to the loss of information and computational costs. To alleviate this problem, transformer is available to model the global dependencies. Bitemporal association, which is described as subtraction or attention mechanism, is not fully considered by current methods. Therefore, we proposed a spatial-temporal association enhanced mobile-friendly vision transformer (STAE-MobileVIT) for change detection of high-resolution images with light weight and high efficiency. On the one hand, a temporal association enhanced MobileVIT block is employed to strengthen the association of bitemporal images during feature extraction. On the other hand, a multiscale feature difference aggregator enhanced by spatial association is designed to fuse semantic and detailed information. Since the lack of binary change detection dataset for BSL, we established a small dataset named BSL-CD, consisting of 1083 pairs of 0.8 m bitemporal images with the size of 256 × 256 pixels, along with the corresponding labels. The experiments on BSL-CD show that our light-weight model surpass seven common methods by 3.48, 5.05, and 1.44 percent on F1, IoU, and OA, which proves the efficiency and accuracy of STAE-MobileVIT.
      PubDate: MON, 23 OCT 2023 09:20:57 -04
      Issue No: Vol. 17, No. null (2023)
       
  • LRDE-Net: Large Receptive Field and Image Difference Enhancement Network
           for Remote Sensing Images Change Detection

    • Free pre-print version: Loading...

      Authors: Lele Li;Liejun Wang;Anyu Du;Yongming Li;
      Pages: 162 - 174
      Abstract: In the field of remote sensing, change detection is a crucial study area. Deep learning has made significant strides in the study of remote sensing image change detection during the past few years. Deep learning techniques still have some drawbacks. The global context cannot be modeled by convolutional neural networks due to the receptive field's restrictions. When extracting visual characteristics, the neural network does not concentrate more on the change region, which results in poor distinction between change and no-change regions. To address these problems, we propose networks with large receptive fields (LRFs) and difference image enhancement. First, we design the LRF strategy. It employs a long kernel shape in one spatial dimension for obtaining a long range of relations. Keeping a narrow kernel size in the other spatial dimension can extract local context information while avoiding interference from irrelevant regions. To focus on the changing features, we design the image difference enhancement (IDE) method, which decreases the distance between invariant features and enlarges the distance between changing features. In addition, we design the cross-channel interaction (CNI) strategy, which models the relationship between feature map channels and extracts feature representations through local CNI. On the CDD, WHU-CD, and LEVIR-CD public datasets, we conducted comprehensive experiments. According to the experimental results, our proposed LRDE-Net performs better than other state-of-the-art change detection techniques, and the change regions are more precisely identified. It can better cope with seasonal changes, light intensity, and other pseudochange disturbances.
      PubDate: MON, 23 OCT 2023 09:20:57 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Swin Transformer Embedding Dual-Stream for Semantic Segmentation of Remote
           Sensing Imagery

    • Free pre-print version: Loading...

      Authors: Xuanyu Zhou;Lifan Zhou;Shengrong Gong;Shan Zhong;Wei Yan;Yizhou Huang;
      Pages: 175 - 189
      Abstract: The acquisition of global context and boundary information is crucial for the semantic segmentation of remote sensing (RS) images. In contrast to convolutional neural networks (CNNs), transformers exhibit superior performance in global modeling and shape feature encoding, which provides a novel avenue for obtaining global context and boundary information. However, current methods fail to effectively leverage these distinctive advantages of transformers. To address this issue, we propose a novel single encoder and dual decoders architecture called STDSNet, which embeds the Swin transformer into the dual-stream network for semantic segmentation of RS imagery. The proposed STDSNet employs the Swin transformer as the network backbone in the encoder to address the limitations of CNNs in global modeling and encoding shape features. The dual decoder comprises two parallel streams, namely the global stream (GS) and the shape stream (SS). The GS utilizes the global context fusion module (GCFM) to address the loss of global context during upsampling. It further integrates GCFMs with skip connections and a multiscale fusion strategy to mitigate large-scale regional object classification errors resulting from similar features or shadow occlusion in RS images. The SS introduces the gate convolution module (GCM) to filter out irrelevant features, allowing it to focus on processing boundary information, which improves the semantic segmentation performance of small targets and their boundaries in RS images. Extensive experiments demonstrate that STDSNet outperforms other state-of-the-art methods on the ISPRS Vaihingen and Potsdam benchmarks.
      PubDate: TUE, 24 OCT 2023 09:17:57 -04
      Issue No: Vol. 17, No. null (2023)
       
  • PSFNet: Efficient Detection of SAR Image Based on Petty-Specialized
           Feature Aggregation

    • Free pre-print version: Loading...

      Authors: Peng Zhou;Peng Wang;Jie Cao;Daiyin Zhu;Qiyuan Yin;Jiming Lv;Ping Chen;Yongshi Jie;Cheng Jiang;
      Pages: 190 - 205
      Abstract: With the rapid development of deep learning, convolutional neural networks have achieved milestones in synthetic aperture radar (SAR) image object detection. However, object detection in SAR images is still a great challenge due to the difficulty in distinguishing targets from complex backgrounds. At the same time, most of the targets in SAR images are small and unevenly distributed, which makes it challenging to extract sufficient feature information. To solve these issues mentioned above, an efficient object detection network for SAR images based on Swin transformer and YOLOv7 is proposed in this article. First, we design a novel feature aggregation module Petty-specialized feature aggregation (PS-FPN) to enrich small targets’ semantic and spatial features while keeping the model lightweight. PS-FPN module uses the fusion of deep and shallow features by using cross-layer feature aggregation and single-branch feature aggregation to enhance the detection of small targets. Second, a novel attention mechanism strategy mix-attention is proposed to find more attention regions. Finally, we add one more prediction head to extract shallow features that effectively preserve small targets’ feature information. To verify the effectiveness of the proposed algorithm, extensive experiments are carried out on several challenging SAR image datasets. The results show that, compared with other state-of-the-art detectors, the proposed method can achieve significant performance based on lightweight detection.
      PubDate: WED, 25 OCT 2023 09:16:38 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Postconstruction Deformation Characteristics of High-Fill Foundations of
           Kunming Changshui International Airport Using Time-Series InSAR Technology
           

    • Free pre-print version: Loading...

      Authors: Mingchun Wen;Mengshi Yang;Xin Zhao;Zhifang Zhao;
      Pages: 206 - 218
      Abstract: The Kunming Changshui International Airport was built on complex mountainous terrain with significant fluctuations in Southwest China. The site contains large areas of high fill susceptible to uneven ground deformation from filling material consolidation, dynamic loading from aircraft, and other factors. Thus, monitoring and analyzing the causes and characteristics of postconstruction deformation of high-fill foundations are crucial to the airport's safe operation. In this study, we obtained a total of 149 Sentinel-1A ascending SAR images from March 2017 to March 2022. We estimated the elevation error and corrected the topographic phase to mitigate the impact of large topographic fluctuations on monitoring. Then, we extracted ground subsidence results based on time-series InSAR technology. The results show that deformation mainly occurred in the high-fill areas during the monitoring period, whereas settlement was more pronounced when the fill height was over 30 m. The deformation rate is influenced by the height of the fill, ground reinforcement measures, and dynamic loading of aircraft. Using the Mann-Kendall trend analysis and Pettitt mutation test methods to detect temporal information from time-series points, we found that the time required for the foundation to reach a stable state after construction is not directly related to its residual subsidence. Other factors, such as construction measures also influence it.
      PubDate: MON, 30 OCT 2023 09:18:20 -04
      Issue No: Vol. 17, No. null (2023)
       
  • ResMorCNN Model: Hyperspectral Images Classification Using
           Residual-Injection Morphological Features and 3DCNN Layers

    • Free pre-print version: Loading...

      Authors: Mohammad Esmaeili;Dariush Abbasi-Moghadam;Alireza Sharifi;Aqil Tariq;Qingting Li;
      Pages: 219 - 243
      Abstract: Hyperspectral imagery is widely used for analyzing substances and objects, specifically focusing on their classification. The advancement of processing capabilities and the emergence of cloud computing platforms have made deep learning (DL) models increasingly popular for accurately and efficiently hyperspectral images (HSI) classification. In addition, utilizing image-processing techniques that employ specific mathematical operations for feature extraction and noise reduction further improves the precision of HSI classification. This study introduces the ResMorCNN model, which utilizes 3-D convolutional layers and morphology mathematics to extract structural information, shapes, and interregional interactions from HSIs. These features are then incorporated into the model's layers using residual connections. This approach significantly enhances the classification accuracy of datasets with different characteristics. In fact, the proposed model achieves an average accuracy higher than the top-performing DL method in a competition. To evaluate the overall effectiveness of the proposed method, it was tested on four distinct and comprehensive datasets, Indian Pines, Pavia University, Houston University, and Salinas. These datasets were carefully selected, taking into account factors such as scale, dispersion, and sample size. The overall accuracy results obtained for each evaluated dataset were 97.81%, 99.33%, 98.67%, and 99.71%, respectively. This demonstrates an average improvement of 3.37% compared to the results of the best-performing method. The results demonstrate the effectiveness of the proposed ResMorCNN model for various applications that require accurate and efficient classification of HSI.
      PubDate: MON, 30 OCT 2023 09:18:20 -04
      Issue No: Vol. 17, No. null (2023)
       
  • MLC30: A New 30 m Land Cover Dataset for Myanmar From 1990 to 2020 Using
           Training Sample Migration Framework

    • Free pre-print version: Loading...

      Authors: Huaqiao Xing;Linye Zhu;Yuqing Zhang;Dongyang Hou;Cansong Li;
      Pages: 244 - 260
      Abstract: Myanmar has experienced rapid socio-economic developments in recent decades, which have a greater impact on land cover change. Accurate long time series land cover datasets for Myanmar can be of great help in environmental protection and natural resource management. However, there are relatively few existing studies on long time series land cover datasets in Myanmar, and the acquisition of training samples within different time series is a big challenge. Therefore, this study used Google Earth Engine and Landsat imagery to produce a land cover dataset for every two years from 1990 to 2020 using a training sample migration framework. First, the differences in index change, spectral value change, and spectral shape change were used to determine whether the sample points had changed between the base year and the previous year, and then a small number of samples were manually selected. Second, the spectral features, index information, and texture information of the remote sensing images and the object-oriented segmentation method were used to obtain object-oriented multidimensional features. Finally, the random forest method was employed to train the samples of the previous year to obtain the land cover data of the previous year. The results of the study show that the average overall precision of the land cover classification results for Myanmar for 1990–2020 is 0.83 and Kappa is 0.79. In addition, the land cover classification results for Myanmar of 1990–2020 are significantly better than those of Globeland30-2020, FROM-GLC, and Dynamic World land cover, and comparing with these products showed good agreement.
      PubDate: MON, 30 OCT 2023 09:18:20 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Spatial Characteristics and Influencing Factors of Night Cultural and
           Tourism Consumption Agglomeration Areas in China

    • Free pre-print version: Loading...

      Authors: Kun Shang;Yuqing Zhang;Xueming Li;Wansheng Li;Guangyu Zhou;
      Pages: 261 - 273
      Abstract: China suffered from severe economic stagnation and urban decay during the COVID-19 pandemic, even public mental health was threatened. To create a favorable public consumption environment for enhancing the economic development, and meeting the needs of public life, the Chinese government has transformed and built a series of national night cultural and tourism consumption agglomeration areas as a new type of functional area in cities. The present study uses spatial analysis methods such as nearest neighbor index, nuclear density analysis, and coefficient of geographic association to quantitatively analyze the spatial distribution characteristics, equilibrium status, distribution density, and internal influential factors. The results showed the following. 1) The spatial distribution of night agglomeration areas in China is unbalanced, with a “rhombus-shaped structures, regional cluster distributions, and single nucleus aggregation points” combination of spatial distribution characteristics. 2) Eight different types of night agglomeration areas with distinctive features, large differences in spatial density, and significant geographical differentiation were identified. 3) The formation of spatial distribution patterns of night agglomeration areas was the result of the joint influence of five factors: resource endowment, economic level, transportation location, guest market, and policy environment. Understanding the current development of night agglomeration areas in China can lay the foundation for future in-depth studies on the spatiotemporal evolution of China's night tourism economy, as well as provide an urban renewal idea and experience for other global countries and regions that are facing economic crises, low urban land use efficiency, and obstruction in promoting new urbanization.
      PubDate: MON, 30 OCT 2023 09:18:20 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Assessing the Freeze/Thaw States in Arctic Circle Using FengYun-3E GNOS-R:
           An Initial Demonstration and Analysis

    • Free pre-print version: Loading...

      Authors: Xuerui Wu;Xinqiu Ouyang;Shengli Wu;Fang Wang;Zheng Duan;
      Pages: 274 - 281
      Abstract: In this article, we present the first demonstration of the FengYun-3E (FY3E) Global Navigation Satellite System Occultation Sounder II-Reflectometry (GNOS-R) payload's capacity to detect near-surface soil freeze/thaw (F/T) states. This study offers an initial analysis of the F/T retrieval algorithm applied to data collected from the Arctic Circle, underscoring the GNOS-R's potential to deliver long-term near-surface soil F/T products. Data for the period extending from the launch day of GNOS-R (Day of Year (DOY) 179, 2021) to DOY 270 in 2022 were analyzed using the surface reflectivity (SR) ratio factor to discriminate F/T variations. Comparisons were made with soil moisture active passive (SMAP) F/T products, serving as an auxiliary analysis. We found a strong consistency between SR ratio factor and SMAP F/T values, with the accuracy of the F/T retrieval algorithm exceeding 60%. These findings corroborate the efficacy of the GNOS-R payload aboard FY3E in monitoring F/T patterns at higher latitudes, specifically, the Arctic Circle. The outcomes of this study will be beneficial for future F/T detection efforts using spaceborne Global Navigation Satellite System-Reflectometry payloads.
      PubDate: MON, 30 OCT 2023 09:18:20 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Deep Learning-Oriented c-GAN Models for Vegetative Drought Prediction on
           Peninsular India

    • Free pre-print version: Loading...

      Authors: Jyoti S. Shukla;Rahul Jashvantbhai Pandya;
      Pages: 282 - 297
      Abstract: In this article, the vegetative drought prediction employing Deep Learning (DL) models is designed, incorporating rainfall data and NOAA satellite-data-derived Vegetation Health Index (VHI) values spanning 1981–2022. Correspondingly, two DL-oriented models based on Generative Adversarial Networks (GANs): 1) Pix2Pix GAN (P2P) and 2) Bidirectional Convolutional LSTM (BiConvLSTM)-P2P GAN (BiCP2P) are developed over the targeted Region of Interest (ROIs). The assimilation of generative DL models for the application of drought forecasting constitutes a novel investigation and a state-of-the-art approach targeted in this work. Subsequently, the primary ROI designated is peninsular India, and the models. efficacy is validated by implementing it on two more ROIs: the Karnataka and Rajasthan states of India. The proposed models. outcomes are compared with several preferred methodologies quantitatively through Coefficient of Determination (R2 score), Mean Squared Error (MSE), and Mean Absolute Error (MAE) and qualitatively employing drought maps denoting the VHI-based drought severity levels over the ROI. Remarkably, excellent performance is demonstrated by the proposed models over peninsular India, with earned R2 score, MSE, and MAE values of 0.971, 0.0016, and 0.020 for P2P and 0.963, 0.0021, and 0.0239 for BiCP2P, respectively. Moreover, generated drought maps efficiently portrayed the drought severities across the land cover and could potentially be extended further for rapid drought risk assessments. The proposed models functioned outstandingly for the developed datasets on the ROIs, corroborating their potential for similar forecasting applications in other climatic zones, which can aid in better planning and preparedness to tackle natural predicaments such as drought calamity.
      PubDate: MON, 30 OCT 2023 09:18:20 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Automated Method for Artificial Impervious Surface Area Mapping in
           Temperate, Tropical, and Arid Environments Using Hyperlocal Training Data
           With Sentinel-2 Imagery

    • Free pre-print version: Loading...

      Authors: Kristofer Lasko;Francis D. O'Neill;
      Pages: 298 - 314
      Abstract: This study presents an automated methodology to generate training data to map up-to-date artificial impervious surface (AIS) extent maps using two dates (winter and nonwinter) of a Sentinel-2 granule across six international sites (Egypt, India, Qatar, U.K., Eastern USA, and Western USA). It uses a series of spectral, textural, and distance decision functions combined with an outdated AIS layer to create nontarget and target binary masks from which to generate a balanced set of training data applied to a random forest classifier. Two outdated global AIS layers (GMIS-2010 and GISA-2016) were evaluated within the framework to create AIS maps from more recent years (e.g., 2020). For the decision functions, stepwise threshold adjustments applied to normalized difference vegetation index (NDVI) and Euclidean distance layers were evaluated on the binary masks (low-density AIS, high-density AIS, and nontarget land covers) with 729 permutations and 115 permutations for global impervious surface area (GISA) and global manmade impervious surface (GMIS), respectively. The optimal thresholds were determined globally (all six scenes), individually (scene) and grouped by climate for adaptive thresholds. The accuracy assessment found both GMIS-output and GISA-output with global thresholds can accurately map current AIS with 86.9% (±1.7%) (GISA) and 82.7% (±2.3%) (GMIS) accuracy. Adaptive climate thresholds yielded slightly higher accuracies for temperate, tropics, and arid scenes. A novel beach bare ground sampling mask and annual NDVI standard deviation were also evaluated for performance and improved the accuracy in 5/6 sites. Lastly, the global GISA output was compared with a manually labeled deep learning model (Esri) with slightly lower overall accuracy (86.9% vs 88.6%).
      PubDate: MON, 30 OCT 2023 09:18:20 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Parallel Computing Method of Commonly Used Interpolation Algorithms for
           Remote Sensing Images

    • Free pre-print version: Loading...

      Authors: Minghu Fan;Xianyu Zuo;Bing Zhou;
      Pages: 315 - 322
      Abstract: Parallel computing is a common method to accelerate remote sensing image processing. This article briefly describes six commonly used interpolation functions and studies three commonly used parallel computing methods of the corresponding nine interpolation algorithms in remote sensing image processing. First, two kinds of general parallel interpolation algorithms (for CPU and GPU, respectively) are designed. Then, in two typical application scenarios (data-intensive and computing-intensive), four computing methods (one serial method and three parallel methods) of these interpolation algorithms are tested. Finally, the acceleration effects of all parallel algorithms are compared and analyzed. On the whole, the acceleration effect of the parallel interpolation algorithm is better in computer-intensive scenario. In CPU-oriented methods, the speedup of all parallel interpolation algorithms mainly depends on the number of physical cores of CPU, whereas in GPU-oriented methods, a speedup is greatly affected by the computation complexity of an algorithm and the application scenario. GPU has a better acceleration effect on the interpolation algorithms with bigger computation complexity and has more advantages in the computing-intensive scenarios. In most cases, GPU-based interpolation is ideal for efficient interpolation.
      PubDate: TUE, 31 OCT 2023 09:16:39 -04
      Issue No: Vol. 17, No. null (2023)
       
  • High-Resolution Remote Sensing Image Zero-Watermarking Algorithm Based on
           Blockchain and SDAE

    • Free pre-print version: Loading...

      Authors: Dingjie Xu;Na Ren;Changqing Zhu;
      Pages: 323 - 339
      Abstract: Existing zero-watermarking algorithms for remote sensing images heavily rely on traditional feature extraction techniques, which are vulnerable to targeted attacks and lack discriminability for images captured by different sensors or at different time periods in the same geographical area. To address these limitations, this article proposes a novel watermarking algorithm based on blockchain and stacked denoising autoencoder (SDAE) to achieve lossless copyright protection for high-resolution remote sensing images. The algorithm utilizes SDAE to extract deep and robust features from local square feature regions for watermark construction. Moreover, the algorithm incorporates a watermark registration scheme designed with Hyperledger Fabric and InterPlanetary file system to ensure secure and trustworthy registration of watermarks and associated parameter information, enhancing the algorithm's uniqueness. Experimental results demonstrate the effectiveness of the proposed algorithm against various watermark attacks and its high discriminability for similar images. This algorithm holds significant potential for wide-ranging applications in the field of lossless copyright protection for high resolution remote sensing images (HRRS), effectively safeguarding the commercial interests of data providers.
      PubDate: TUE, 31 OCT 2023 09:16:39 -04
      Issue No: Vol. 17, No. null (2023)
       
  • How Does Super-Resolution for Satellite Imagery Affect Different Types of
           Land Cover' Sentinel-2 Case

    • Free pre-print version: Loading...

      Authors: Anna Malczewska;Maciej Wielgosz;
      Pages: 340 - 363
      Abstract: In the dynamic field of satellite imagery, the significance of super-resolution (SR) techniques, grounded on advanced deep learning methods, is paramount. A thorough understanding and remediation of the distinct challenges posed by various land cover types for image resolution enhancement form the essence of this research. This work diligently employs two unique neural networks, SRCNN and SwinIR Transformer, to scrutinize their varying impacts on a range of land cover types, ensuring a detailed and comprehensive exploration. This study transcends the mere enhancement of the Sentinel-2 dataset's resolution from 20 m/pix to 10 m/pix. It ambitiously seeks to excavate the intricate trends inherent to different land cover types and their corresponding interactions with SR processes. The application of neural networks on 255 × 254 pixel patches, covering six dominant types—forests, large fields, small fields, urban, sub-urban, and mixed—highlights substantial variations in metrics, underlining the individual interactions of each land cover type with SR techniques. A comprehensive accuracy assessment is meticulously conducted, employing an array of metrics and frequency domains to shed light on the nuanced differences and provide vital insights for optimizing each land cover type's SR approaches. Notably, the PSNR metric reveals significant disparities, particularly in the “forest” and “urban” categories for both SRCNN and SwinIR. According to the PSNR metric, the “forest” class yielded the best results with 66.06 for SRCNN and 67.00 for SwinIR, while the “urban” class marked the lowest with 55.09 and 57.02, respectively, reinforcing the critical nature of this study.
      PubDate: TUE, 31 OCT 2023 09:16:39 -04
      Issue No: Vol. 17, No. null (2023)
       
  • TKP-Net: A Three Keypoint Detection Network for Ships Using SAR Imagery

    • Free pre-print version: Loading...

      Authors: Xiunan Li;Peng Chen;Jingsong Yang;Wentao An;Gang Zheng;Dan Luo;Aiying Lu;Zimu Wang;
      Pages: 364 - 376
      Abstract: Remote-sensing ship monitoring is a crucial area of research with key applications in military and civilian fields. The ability to extract information, such as ship length, width, and heading from remote-sensing data, particularly from synthetic aperture radar (SAR) images, is of paramount importance. Current state-of-the-art SAR image ship monitoring focuses primarily on ship detection. Assessing the direction of ships usually relies on the observability of wake features. However, the observability of these wake features is often affected by factors, such as the SAR system parameters, ship attributes, and dynamic marine environments. This can make accurate direction assessments a challenging task. In response to these challenges, this study has presented a novel and effective algorithm for ship monitoring from SAR images based on an anchor-free framework and the powerful feature extraction capabilities of convolutional neural networks. The proposed method learned the scattering and morphological information of a ship's bow and stern from high-resolution SAR images to determine the ship's direction with a high level of accuracy using a rotating bounding box. The algorithm was tested on a dataset, achieving an average precision of 90.8% and bow classification accuracy of 92.5%, demonstrating its potential contributing to the advancement of remote sensing.
      PubDate: WED, 01 NOV 2023 09:17:45 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Study on Four-Dimensional Evolution of Concentric Traveling Ionospheric
           Disturbances After Falcon 9 Rocket Launch Using Ionospheric Tomography

    • Free pre-print version: Loading...

      Authors: Yutian Chen;Dongjie Yue;Changzhi Zhai;
      Pages: 377 - 387
      Abstract: The ionospheric disturbances induced by the Falcon 9 rocket launch on 17 January 2016 were reconstructed by three-dimensional computerized ionospheric tomography using observations from the North American global navigation satellite system networks. The results showed that concentric traveling ionospheric disturbances (CTIDs) occurred ∼18 min after the rocket launch and were remarkable at 200–300 km altitudes. The vertical phase velocities of the CTIDs were consistent with the inclinations of the U-shaped structures. At specific azimuth directions of 350°, 30°, and 105°, the estimated vertical phase velocities between 100 and 200 km altitudes were ∼222.2, ∼208.3, and ∼242.4 m/s, respectively. When the CTIDs propagated upward to 400–500 km altitudes, their vertical velocities increased to ∼566.7, ∼966.7, and ∼944.4 m/s. CTIDs traveling northward (azimuths 350°, 30°) had periods of ∼11 min. At 200 and 300 km altitudes, their horizontal phase velocities were ∼309.9–323.3 and ∼309.4–330.9 m/s, respectively, with horizontal wavelengths of ∼204.5–213.4 and ∼204.2–218.4 km. In contrast, CTIDs propagating eastward (azimuth 105°) displayed a period of ∼15 min. At 200 and 300 km altitudes, their horizontal phase velocities were ∼223.2 and ∼241.1 m/s, respectively, with horizontal wavelengths of ∼200.9 and ∼217.0 km. These CTIDs propagation characteristics agreed well with the theory of gravity waves.
      PubDate: WED, 01 NOV 2023 09:17:43 -04
      Issue No: Vol. 17, No. null (2023)
       
  • CSPPartial-YOLO: A Lightweight YOLO-Based Method for Typical Objects
           Detection in Remote Sensing Images

    • Free pre-print version: Loading...

      Authors: Siyu Xie;Mei Zhou;Chunle Wang;Shisheng Huang;
      Pages: 388 - 399
      Abstract: Detecting and recognizing objects are crucial steps in interpreting remote sensing images. At present, deep learning methods are predominantly employed for detecting objects in remote sensing images, necessitating a significant number of floating-point computations. However, low computing power and small storage in computing devices are hard to afford the large model parameter quantity and high computing complexity. To address these constraints, this article presents a lightweight detection model called CSPPartial-YOLO. This model introduces the partial hybrid dilated convolution (PHDC) Block module that combines hybrid dilated convolutions and partial convolutions to increase the receptive field at a low computational cost. By using the PHDC Block within the model design framework of cross-stage partial connection, we construct CSPPartialStage that reduces computational burden without compromising accuracy. Coordinate attention module is also employed in CSPPartialStage to aggregate position information and improve the detection of small objects with complex distributions in remote sensing images. A backbone and neck are developed with CSPPartialStage, and the rotation head of the PPYOLOE-R model adapts to objects of multiple orientations in remote sensing images. Empirical experiments using the dataset for object deTection in aerial images (DOTA) dataset and a large-scale small object detection dAtaset (SODA-A) dataset indicate that our method is faster and resource efficient than the baseline model (PPYOLOE-R), while achieving higher accuracy. Furthermore, comparisons with current state-of-the-art YOLO series detectors show our proposed model's competitiveness in terms of speed and accuracy. Moreover, compared to mainstream lightweight networks, our model exhibits better hardware adaptability, with lower inference latency and higher detection accuracy.
      PubDate: WED, 01 NOV 2023 09:17:43 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Uncertainty Budget for a Traceable Operational Radiometric Calibration of
           Field Spectroradiometers, Calibration of the Heliosphere

    • Free pre-print version: Loading...

      Authors: Mike Werfeli;Andreas Hueni;Dessislava Ganeva;Giulia Ghielmetti;Laura Mihai;
      Pages: 400 - 410
      Abstract: To measure the distinct interaction of the Earth's materials with solar electromagnetic radiation, field spectroradiometers are commonly utilized. These are used to validate spectroradiometers deployed on various platforms through comparison exercises. Following metrology standards, the inclusion of uncertainties is required. Thus, field spectroradiometers need to be calibrated regularly against traceable radiance sources. In this article, we present a laboratory radiometric calibration protocol for the calibration of a heliosphere integrating sphere to make it traceable to the International System of Units as well as to establish an uncertainty budget. We adopted a transfer radiometer approach including four spectroradiometers that were calibrated at the Deutsches Zentrum für Luft und Raumfahrt Radiometric Standard facility before transferring that calibration to the heliosphere. After considering various sources of uncertainty by employing an uncertainty tree diagram approach, we arrive at an overall propagated uncertainty of approximately 1.5%. In future publications, we will present how to extend the traceability to other attenuations provided by the heliosphere. Its application to the calibration of a field spectroradiometer will be the focus of a future publication.
      PubDate: FRI, 03 NOV 2023 09:16:47 -04
      Issue No: Vol. 17, No. null (2023)
       
  • ABLAL: Adaptive Background Latent Space Adversarial Learning Algorithm for
           Hyperspectral Target Detection

    • Free pre-print version: Loading...

      Authors: Long Sun;Zongfang Ma;Yi Zhang;
      Pages: 411 - 427
      Abstract: Hyperspectral images (HSIs) are challenging for hyperspectral object detection (HTD) due to their complex background information and the limited prior knowledge of the target. This article proposes an adaptive background latent space adversarial learning algorithm for hyperspectral target detection (ABLAL). We begin by using a coarse screening method to select pseudobackground and pseudotarget sample sets, addressing the issues caused by insufficient prior target information and complicated background information, which result in low detection accuracy. Next, we utilize an Adversarial Autoencoder (AAE) based backbone network to extract the background latent spatial information of the HSI. It should be noted that we adaptively constrain the accuracy of the extracted information through the pseudotarget dataset, accounting for the impact of potential targets in the pseudobackground dataset. Furthermore, we fully utilize the information extracted by AAE and employ a strategy combining multiple output results of AAE. Specifically, we use the distance between the target latent space vector and the background latent space vector, and the HSI reconstruction difference to suppress the background. Finally, extensive experiments are conducted on real datasets to demonstrate the effectiveness of the proposed method.
      PubDate: FRI, 03 NOV 2023 09:16:47 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Sea Surface Signal Extraction for Photon-Counting LiDAR Data: A General
           Method by Dual-Signal Unmixing Parameters

    • Free pre-print version: Loading...

      Authors: Zhen Wen;Xinming Tang;Guoyuan Li;Bo Ai;Guanghui Wang;Jiaqi Yao;Fan Mo;
      Pages: 428 - 437
      Abstract: The ice, cloud, and land elevation satellite-2 (ICESat-2) is the only satellite that produces photon-counting light detection and ranging data, and is equipped with the advanced topographic laser altimeter system. ICESat-2 provides sea surface height product; however, its approach of the product is unsuitable for areas with sub-surface signals. Conventional denoising methods applied to sea surface photon data of variable density involve the use of different empirical parameters. Considering the distribution of sea surface signal photons, we propose a general open-source method using a dual-signal unmixing parameter (DSUMP), which incorporates the Gaussian distribution of dual-signal peaks to determine the sea surface range. This method facilitates the direct extraction of sea surface photons under various observation conditions—day or night, strong or weak beams, and including or excluding seabed photons—without requiring any variable parameters. The elevation error by DSUMP within 0.1m accounts for more than 97%. The mean absolute error is within 0.01 m compared to sea surface photons obtained via manual extraction. Different model parameters show stable denoising accuracy, only affects operating efficiency. The proposed method introduces a novel denoising technique for extracting sea surface elevation from ICESat-2 altimetry data, and its applicability can be extended to various point cloud data with similar distributions.
      PubDate: FRI, 03 NOV 2023 09:16:47 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Extraction Methods for Small-Scale Features on a Large Scale:
           Investigating Object-Oriented Cart Decision Tree for Gravel Information
           Extraction

    • Free pre-print version: Loading...

      Authors: Yuxin Chen;Weilai Zhang;Jiajia Yang;Yuanyuan Xu;Qian Yuan;Wuxue Cheng;Li Peng;
      Pages: 438 - 449
      Abstract: This study employs the object-oriented Cart (Classification and Regression Trees) decision tree methodology to delineate contiguous gravel areas within Zamu town. The primary objective is to devise a technique capable of efficiently identifying small-scale features on a macroscopic scale. Given the pervasive and uninterrupted distribution of background elements like forests, snow, and water in the designated study zone, the removal of these features can significantly bolster the precision of target feature extraction. The elimination of these background elements predominantly hinges on the application of index thresholds. Specifically, the Normalized Difference Vegetation Index (NDVI), Normalized Difference Water Index (NDWI), and Normalized Difference Snow Index (NDSI) are employed to filter out these nontarget features. Following this, Google HD historical imagery is integrated with Sentinel-2 data to facilitate the object-oriented Cart decision tree-based feature extraction. Our findings underscore that both NDVI and NDWI are pivotal in eradicating forest backgrounds. For differentiating snow-covered terrains, the NDVI and NDSI indices prove particularly vital. The identification of water bodies necessitates the synergistic use of all three indices. Notably, the Cart decision tree approach, grounded in the “cull, filter, classify, and merge” philosophy, showcases superior classification accuracy relative to other supervised classification techniques. In the realm of decision tree rule formulation, spectral features dominate, constituting 47.6% of the land class classification. Concurrently, texture features are instrumental, accounting for 38.1%. These texture features exhibit an enhanced discriminatory capacity, whereas the incorporation of diverse indices offers limited incremental value. Pertinently, within the suite of features conducive to gravel extraction, both the Blue band and gray-level co-occurrence matrix entropy emerge as particularly efficacious.
      PubDate: FRI, 03 NOV 2023 09:16:47 -04
      Issue No: Vol. 17, No. null (2023)
       
  • A Novel Method for Identifying Crops in Parcels Constrained by
           Environmental Factors Through the Integration of a Gaofen-2
           High-Resolution Remote Sensing Image and Sentinel-2 Time Series

    • Free pre-print version: Loading...

      Authors: Weijia Chen;Guilin Liu;
      Pages: 450 - 463
      Abstract: Accurately mapping crop cultivation types is essential for the sustainable development of precision agriculture. Environmental restrictions on crop growth, such as soil salinization in arid zones, generally lead to spatial crop growth heterogeneity within cropland fields, which in turn generates differences in the spectral responses reflected in optical remote sensing images of the same croplands and leads to pixel-scale crop-mapping misclassifications. Thus, through this article, we proposed a method to solve this problem at the geoparcel scale by integrating geometric features from a Gaofen-2 high-resolution remote sensing image and the spectral-temporal features derived from Sentinel-2 time series. The results showed that cropland parcels could be accurately extracted from Gaofen-2 images by employing the U-Net semantic segmentation model with an overall accuracy (OA) reaching 97% and a kappa coefficient of 0.95. Then, geoparcel-scale crop types were mapped based on prior crop phenology knowledge and the corresponding Sentinel-2 time series using the time-weighted dynamic time warping (TWDTW) classification algorithm. The parcel-based TWDTW algorithm had an OA of 99.64%, a kappa coefficient of 0.99, and optimal spatial homogeneity in the results, thus outperforming the pixel-based TWDTW method. These results provide a potential solution for mapping crops under spatially heterogeneous cropland conditions affected by various environmental constraints.
      PubDate: FRI, 03 NOV 2023 09:16:47 -04
      Issue No: Vol. 17, No. null (2023)
       
  • UAVPal: A New Dataset for Semantic Segmentation in Complex Urban Landscape
           With Efficient Multiscale Segmentation

    • Free pre-print version: Loading...

      Authors: Abhisek Maiti;Sander Oude Elberink;George Vosselman;
      Pages: 464 - 475
      Abstract: Semantic segmentation has recently emerged as a prominent area of interest in Earth observation. Several semantic segmentation datasets already exist, facilitating comparisons among different methods in complex urban scenes. However, most open high-resolution urban datasets are geographically skewed toward Europe and North America, while coverage of Southeast Asia is very limited. The considerable variation in city designs worldwide presents an obstacle to the applicability of computer vision models, especially when the training dataset lacks significant diversity. On the other hand, naively applying computationally expensive models leads to inefficacies and sometimes poor performance. To tackle the lack of data diversity, we introduce a new UAVPal dataset of complex urban scenes from the city of Bhopal, India. We complement this by introducing a novel dense predictor head and demonstrate that a well-designed head can efficiently take advantage of the multiscale features to enhance the benefits of a strong feature extractor backbone. We design our segmentation head to learn the importance of features at various scales for each individual class and refine the final dense prediction accordingly. We tested our proposed head with a state-of-the-art backbone on multiple UAV datasets and a high-resolution satellite image dataset for LULC classification. We observed improved intersection over union (IoU) in various classes and up to 2$\%$ better mean IoU. Apart from the performance improvements, we also observed nearly 50$\%$ reduction in computing operations required when using the proposed head compared to the traditional segmentation head.
      PubDate: MON, 06 NOV 2023 09:18:29 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Complex Permittivity Retrieval Approach With Radar Enhanced Contrast
           Source Inversion for Microwave Nondestructive Road Evaluation

    • Free pre-print version: Loading...

      Authors: Katsuyoshi Suzuki;Shingo Nakamura;Shouhei Kidera;
      Pages: 476 - 488
      Abstract: In this article, an experimental investigation of microwave quantitative imaging of nondestructive testing (NDT) applications for concrete road structures is presented. A radar prior-based contrast source inversion (CSI) scheme is introduced to achieve an accurate reconstruction for target shape and dielectric property by applying the promising radar approach–range points migration. Moreover, to address the local optimum problem, we introduce an appropriate initial estimate for complex permittivity that focuses on the cost function of the CSI. A numerical and real data from an actual concrete road model, where a thin-layer water-filled cavity is buried into the boundary area between the asphalt and the concrete floorboard, show that our proposed approach improves reconstruction accuracy for permittivity and conductivity in a real NDT model.
      PubDate: WED, 08 NOV 2023 09:16:34 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Permuted KPCA and SMOTE to Guide GAN-Based Oversampling for Imbalanced HSI
           Classification

    • Free pre-print version: Loading...

      Authors: Tajul Miftahushudur;Bruce Grieve;Hujun Yin;
      Pages: 489 - 505
      Abstract: Lack of sufficient and balanced data is one of the major challenges in hyperspectral image classification. This problem can cause poor classification performance, especially in detecting or classifying samples of minority classes. The easiest way to overcome the problem is by resampling or creating synthetic samples to balance the class distributions. As the most advanced generative method, generative adversarial networks (GANs) have been used for generating synthetic data. However, GANs need a large amount or sufficient minority class data to train. In this article, we propose to leverage the synthetic minority oversampling technique (SMOTE) in GANs for creating high quality synthetic data to tackle the imbalance problem. The main idea is to train the generator of the GAN to synthesize data from pattern vectors instead of random noise vectors so to guide the GAN to produce data that can expand the minority class data on the decision boundaries. We used kernel principal component analysis and SMOTE to create the pattern vectors and used a silhouette score to control and prevent overlapping issues. In addition, we applied a self-attention module and an automatic data filter to further minimize potentially wrongly labeled or overlapping samples before being added into the training set. Experimental results on both hyperspectral and remote sensing datasets show that the proposed technique can generate more realistic, diverse, and unambiguous synthetic data, resulting in significantly improved classification performances over the existing oversampling techniques.
      PubDate: WED, 08 NOV 2023 09:16:34 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Multilateral Semantic With Dual Relation Network for Remote Sensing Images
           Segmentation

    • Free pre-print version: Loading...

      Authors: Weiheng Zhao;Jiannong Cao;Xueyan Dong;
      Pages: 506 - 518
      Abstract: Semantic segmentation of remote sensing images is an extensively employed and demanding task. Although deep convolutional neural networks have significantly increased the accuracy of semantic segmentation, the problems of losing detailed features in segmentation and ignoring rich contextual information of images still exist. To solve these challenges, we propose a multilateral semantic with dual relation network (MSDRNet) for remote sensing images segmentation. The proposed MSDRNet consists of two parallel modules, the detail semantic module and the global semantic module, for extracting image detail and global features, respectively. Subsequently, improved spatial relation block and channel relation block are introduced in two separate parallel modules to further enhance the contextual connection of the images. Finally, a feature refinement module is added to balance the multilateral features between the features extracted from the two branches. We display the robustness and effectiveness of the proposed MSDRNet on the publicly available ISPRS Potsdam and Vaihingen datasets. We further experimented with the Gaofen image dataset, which contains information on larger scale features, to demonstrate the validity of our model. The results of extensive experiments conducted on the aforementioned three datasets show that the proposed approach outperforms several state-of-the-art semantic segmentation methods.
      PubDate: WED, 08 NOV 2023 09:16:34 -04
      Issue No: Vol. 17, No. null (2023)
       
  • MDBES-Net: Building Extraction From Remote Sensing Images Based on
           Multiscale Decoupled Body and Edge Supervision Network

    • Free pre-print version: Loading...

      Authors: Shengjun Xu;Miao Du;Yuebo Meng;Guanghui Liu;Jiuqiang Han;Bohan Zhan;
      Pages: 519 - 534
      Abstract: The extraction of buildings in aerial remote sensing applications is an important and challenging task. Most existing methods extract buildings based on local area attention, ignoring the loss of accuracy due to the global structure of the building. However, global structural features of buildings with strong coupling relationships in complex scenes are difficult to extract, such as the edges and bodies of buildings, leading to discontinuous results. Therefore, multiscale decoupled body and edge supervision network (MDBES-Net), which can consider both edge optimization and inner consistency, is proposed to solve these problems. MDBES-net consists of the body-mask-edge consistency constraint base network (BMECC), decoupling the body and edge aware module (DBEA), and the channel decoupled attention module (CDA). First, body-mask-edge consistency constraint supervision is established by body and edge labels to jointly improve the segmentation effect in the BMECC base network. Second, in the mutiscale DBEA module, building features are warped by a learnable flow field to make body parts more consistent and edges more detailed. Finally, the CDA module performs adaptive calibration of the recoupled feature map channel response to minimize external background noise interference. Experiments on the open Massachusetts building dataset, WHU Building Dataset show that the proposed MDBES-Net can accurately extract buildings in complex scenarios, enabling complete building segmentation with refined boundaries and improved internal consistency.
      PubDate: THU, 09 NOV 2023 09:16:24 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Unveiling Roadway Hazards: Enhancing Fatal Crash Risk Estimation Through
           Multiscale Satellite Imagery and Self-Supervised Cross-Matching

    • Free pre-print version: Loading...

      Authors: Gongbo Liang;Janet Zulu;Xin Xing;Nathan Jacobs;
      Pages: 535 - 546
      Abstract: Traffic accidents threaten human lives and impose substantial financial burdens annually. Accurate estimation of accident fatal crash risk is crucial for enhancing road safety and saving lives. This article proposes an innovative approach that utilizes multiscale satellite imagery and self-supervised learning for fatal crash risk estimation. By integrating multiscale imagery, our network captures diverse features at different scales, encompassing observations of surrounding environmental factors in low-resolution images that cover larger areas and learning detailed ground-level information from high-resolution images. One advantage of our work is its sole reliance on satellite imagery data, making it an efficient and practical solution, especially when other data modalities are unavailable. With the ability to accurately estimate fatal crash risk, our method exhibits a potential for enhancing road safety, optimizing infrastructure planning, preventing accidents, and ultimately saving lives.
      PubDate: THU, 09 NOV 2023 09:16:24 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Trailing Edge Slope Ocean Wind Speed Retrievals Using CYGNSS's Full DDMs

    • Free pre-print version: Loading...

      Authors: Stephen J. Katzberg;Mohammad Al-Khaldi;Faozi Said;Jeonghwan Park;
      Pages: 547 - 556
      Abstract: The launch of the CYGNSS constellation in 2016 represents NASA's first mission dedicated to the acquisition of GPS signals reflected off the earth’s surface. The major geophysical measurement objective is ocean surface winds with particular emphasis placed on the improved characterization of tropical cyclones. The conventional method for retrieving ocean surface winds using spaceborne Global Navigation Satellite System Reflectometry systems, of which CYGNSS is one example, is to measure the reflected power from the specular point on the ocean surface. Accurate measurement of surface power is complicated by calibration of the various power levels from the transmitting satellites, geometric effects, receiving system RF chain characteristics, among others. This article applies an alternative trailing edge slope (TES) method, which is more immune to end-to-end calibration uncertainties. It is nonetheless noted that because TES is a fundamentally shape-based measure of the waveform, it is susceptible to factors that may distort the measurements' slopes. This included measurements made with low SNR (on the order of 1.5 dB or less) due to low level of surface scattering or observations within the low gain portion of the receive antenna pattern. For this reason, in this work retrieval attempts are largely limited to those with a receive gain on the order of 10 dB or better. The TES method is based on the slope of the measured signal as a function of delay. Given the need for a delay space that extends beyond the $\approx$2.5–3.5 chips provided by CYGNSS's standard Level-1 measurements, its utility is explored using one of the constellation's special “full DDM” downlink modes. The TES method is shown to provide competitive retrievals including measurements from the surface, where the conventional method provides ocean wind speed estimates of lower accuracy. Tradeoffs using the TES method are discussed as well as simple improvements in the method are given. It is also noted that because with TES, measurements' slopes are the fundamental observable no calibration of measurements is required.
      PubDate: MON, 13 NOV 2023 09:17:22 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Assimilation of Earth Observation Data for Crop Yield Estimation in
           Smallholder Agricultural Systems

    • Free pre-print version: Loading...

      Authors: Biniam Sisheber;Michael Marshall;Daniel Mengistu;Andrew Nelson;
      Pages: 557 - 572
      Abstract: Crop yield estimates are an important data output of agricultural monitoring systems. In sub-Saharan Africa, large input requirements of crop growth models, fragmented agricultural systems, and small field sizes are substantial challenges to accurately estimate crop yield. Multisensor data fusion can be a valuable source of high spatial and temporal resolution data to meet the requirements of crop growth models in Africa. In this study, we estimated crop yield in smallholder agricultural systems of Ethiopia by assimilating Landsat and MODIS fused data in the simple algorithm for crop yield estimation (SAFY) model. Enhanced spatial and temporal adaptive reflectance fusion model (ESTARFM) adapted for fragmented agricultural landscapes was used for data fusion. We assimilated LAI and phenology information derived from Landsat–MODIS fusion, MODIS and field data for comparison. The model was validated with in situ LAI and yield measured in rice and maize fields during the 2019 growing season. Data fusion minimized the yield estimation error (rRMSE = 16% for maize and rRMSE = 23% in rice) more than MODIS (rRMSE = 20% for maize and rRMSE = 35% in rice) because of its higher LAI and phenology estimation accuracy. Data fusion improved the calibration accuracy of the field and crop-specific model parameters and better captured the spatial variability of yield, which is vital for crop production monitoring and food security in smallholder agricultural systems in Africa. Considering the promising results, further investigation into the transferability of the approach to other smallholder agricultural landscapes and hybridization with machine learning is needed for large-area applications.
      PubDate: WED, 01 NOV 2023 09:17:43 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Recent Ground Displacement Over Permafrost in Midwestern Spitsbergen,
           Svalbard: InSAR Measurements and Modeling

    • Free pre-print version: Loading...

      Authors: Yining Yu;Yu Zhou;Fengming Hui;Xiao Cheng;Kang Wang;
      Pages: 573 - 583
      Abstract: Quantifying the characteristics of ground displacement is of great importance for understanding the geological, hydrological, and biochemical processes occurring due to widespread permafrost degradation. This study investigated the dynamics of annual and seasonal ground displacement over permafrost in midwestern Spitsbergen, Svalbard, using radar interferometry (InSAR) and a permafrost thermal model. Permafrost ground displacement exhibited an evident and widespread subsidence at a rate of ∼2.0 mm/yr during 2018–2021. A composite thermal index model was applied to simulate thawing-season ground displacement in four in situ boreholes based on the in situ observations and the ERA5 Land reanalysis temperature, respectively. The modeled seasonal ground subsidence demonstrated a similar pattern with InSAR measurement (correlation coefficient: 0.86–0.91), indicating the practicability of employing reanalysis temperature in large-scale ground subsidence modeling over data-sparse permafrost regions. The effects of environmental driving factors, including temperature and landcover, on ground displacement were also discussed. This study confirmed the availability of integrating InSAR measurement and thermal-modeled results toward a comprehensive understanding of ground displacement over permafrost regions.
      PubDate: MON, 13 NOV 2023 09:17:22 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Adjustment of Sentinel-3 Spectral Bands With Sentinel-2 to Enhance the
           Quality of Spatio-Temporally Fused Images

    • Free pre-print version: Loading...

      Authors: Meryeme Boumahdi;Angel García-Pedrero;Mario Lillo-Saavedra;Consuelo Gonzalo-Martin;
      Pages: 584 - 600
      Abstract: Spatiotemporal fusion (STF) methods are a paramount solution for generating high spatial and temporal time series, overcoming the limitations of spatial and temporal resolution of satellite data. STF methods typically rely on band-by-band fusion, assuming spectral similarities. However, selecting the optimal band for fusion becomes challenging when multiple narrow bands overlap with the target band, often leading to the use of only one single band. Furthermore, sensor specifications and observation configurations can further compound this challenge, reducing spectral and spatial information. We introduce a new preprocessing step that maximizes the use of spectral information from narrow bands. It minimizes radiometric differences caused by sensor variations in the STF process by considering the spectral response function (SRF). Our method generates adjusted bands that closely match the target band's spectral characteristics, leveraging all available spectral information. We evaluated this strategy at two study sites employing Sentinel 2 and Sentinel 3 data by comparing fused images from adjusted bands and the original bands using three popular STF methods. The results obtained showed that the images fused with the adjusted bands were closer to the target images and achieved better performance, improving the fusion quality compared to the original bands (SAM by 37% and RMSE by 30%). The preprocessing step offers a feasible approach to generate spectral bands that would be captured by the sensors if they had the same spectral characteristics. Importantly, this preprocessing technique is applicable to any STF method.
      PubDate: WED, 15 NOV 2023 09:16:35 -04
      Issue No: Vol. 17, No. null (2023)
       
  • A Novel Phase Unwrapping Method for Low Coherence Interferograms in Coal
           Mining Areas Based on a Fully Convolutional Neural Network

    • Free pre-print version: Loading...

      Authors: Yu Yang;Bingqian Chen;Zhenhong Li;Chen Yu;Chuang Song;Fengcheng Guo;
      Pages: 601 - 613
      Abstract: Subsidence caused by underground coal mining activities seriously threatens the safety of surface buildings, and interferometric synthetic aperture radar has proven to be one effective tool for subsidence monitoring in mining areas. However, the environmental characteristics of mining areas and the deformation behavior of mining subsidence lead to low coherence of interferogram. In this case, traditional phase unwrapping methods have problems, such as low accuracy, and often fail to obtain correct deformation information. Therefore, a novel phase unwrapping method is proposed using a channel-attention-based fully convolutional neural network (FCNet-CA) for low coherence mining areas, which integrates multiscale feature extraction block, bottleneck block, and can better extract interferometric phase features from the noise. In addition, based on the mining subsidence prediction model and transfer learning method, a new sample generation strategy is proposed, making the training dataset feature information more diverse and closer to the actual scene. Simulation experiment results demonstrate that FCNet-CA can restore the deformation pattern and magnitude in scenarios with high noise and fringe density (even if the phase gradient exceeds π). FCNet-CA was also applied to the Shilawusu coal mining area in Inner Mongolia Autonomous Region, China. The experimental results show that, compared with the root mean square error (RMSE) of phase unwrapping network and minimum cost flow, the RMSE of FCNet-CA in the strike direction is reduced by 67.9% and 29.5%, respectively, and by 72.4% and 50.9% in the dip direction, respectively. The actual experimental results further verify the feasibility and effectiveness of FCNet-CA.
      PubDate: WED, 15 NOV 2023 09:16:35 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Transformer-Based Dual-Branch Multiscale Fusion Network for Pan-Sharpening
           Remote Sensing Images

    • Free pre-print version: Loading...

      Authors: Zixu Li;Jinjiang Li;Lu Ren;Zheng Chen;
      Pages: 614 - 632
      Abstract: Due to the limitations of satellite sensors, we can only obtain MS images and PAN images separately. The focus of our attention is to utilize the pan-sharpening method to generate the high-resolution multispectral (HRMS) images. In this article, we proposed the dual-branch multiscale fusion network, which based on the spatial-spectral transformer to comprehensively capture the information contained in MS images and PAN images at different scales. The architecture of our network consists of three parts: during the feature extraction and image fusion stage, we first independently apply upscaling and downscaling operations to the MS and PAN images. Subsequently, we concatenate the images from the two distinct branches and input them into the shallow feature extraction module individually. And then we input them into our adaptive feature extraction block to further extract the crucial details of the images using the attention mechanism. The images a various scales in different branches are then passed through three spectral transformer and three spatial transformer modules to perform a comprehensive extraction of both spatial and spectral characteristics. Finally, the residual local feature module is utilized during the image reconstruction part to deeply extract intricate information from the images and obtain the final HRMS fused image. We have conducted both simulated and real experiments on the benchmark datasets QB and WV2. The final qualitative and quantitative comparative results demonstrate that our innovative method outperforms the current SOTA methods.
      PubDate: MON, 13 NOV 2023 09:17:21 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Comparison Between Thermal-Optical and L-Band Passive Microwave Soil
           Moisture Remote Sensing at Farm Scales: Towards UAV-Based Near-Surface
           Soil Moisture Mapping

    • Free pre-print version: Loading...

      Authors: Nan Ye;Jeffrey P. Walker;Ying Gao;Ivan PopStefanija;James Hills;
      Pages: 633 - 642
      Abstract: The unmanned aerial vehicle (UAV) based remote sensing has drawn increased attention in precision agriculture. Lightweight optical and thermal sensors have been used widely on UAVs for a range of applications, and have been proposed by some as the best approach to map soil moisture at farm scales. However, passive microwave remote sensing has been widely acknowledged as the most accurate soil moisture mapping technology, and adopted by the soil moisture and ocean salinity and soil moisture active and passive satellite missions. Accordingly, it is postulated that this will also be the best technique for UAV-based near-surface soil moisture remote sensing, overcoming the spatial resolution limitation from low earth orbit altitude. Being so far limited by sensor availability, only a small number of studies have illustrated the potential of UAV-based near-surface soil moisture mapping using L-band microwave radiometers, and there has been no direct comparison with the thermal-optical alternative. To guide the design of future UAV-based soil moisture mapping systems, airborne optical, thermal infrared, and passive microwave observations collected from a scientific aircraft at low altitude over a center-pivot irrigation farm in Tasmania, Australia were used in this study to simulate UAV-based observations, and the performances of the thermal-optical and microwave techniques when compared at 75 m scale. The L-band microwave emission showed a superior sensitivity to near-surface soil moisture, and a higher and more consistent soil moisture retrieval accuracy than thermal-optical, with a root-mean-squared error of 0.05–0.06 m3/m3 and 0.05–0.09 m3/m3, respectively.
      PubDate: TUE, 31 OCT 2023 09:16:39 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Toward an Operational Monitoring of Oak Dieback With Multispectral
           Satellite Time Series: A Case Study in Centre-Val De Loire Region of
           France

    • Free pre-print version: Loading...

      Authors: Florian Mouret;David Morin;Hilaire Martin;Milena Planells;Cécile Vincent-Barbaroux;
      Pages: 643 - 659
      Abstract: This article studies the monitoring of oak dieback in forests of the Centre-Val de Loire region (France), where drought-induced dieback has become a major concern due to climate change. The main objective of the study is to evaluate the applicability of multispectral satellite time series for operational monitoring of forest dieback. Using in situ data collected from 2017 to 2022 on approximately 2700 oak plots, a multiyear mapping of the analyzed region was performed using the random forest algorithm and Sentinel-2 images. Our results show that it is possible to detect oak dieback accurately (average overall accuracy = 80% and average balanced accuracy = 79%). A spatial cross-validation analysis also evaluates the performance of the model on regions that were never encountered during training, across all years, resulting in a slight decrease in accuracy ($\sim$5%). The study also highlights the importance of measuring the stability and performance of the classification model over time, in addition to standard cross-validation metrics. A feature analysis shows that the shortwave infrared part of the spectrum is the most important for mapping forest dieback, while the red-edge portion of the spectrum can increase the stability of the model over time. Overall, both in situ data and model predictions showed evidence of forest decline in many areas of the study region. Our results suggest that large areas of forest can decline over short periods of time, highlighting the interest of satellite data to provide timely and accurate information on forest status.
      PubDate: MON, 13 NOV 2023 09:17:22 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Potential of GNSS-R for the Monitoring of Lake Ice Phenology

    • Free pre-print version: Loading...

      Authors: Yusof Ghiasi;Claude R. Duguay;Justin Murfitt;Milad Asgarimehr;Yuhao Wu;
      Pages: 660 - 673
      Abstract: This article introduces the first use of global navigation satellite system (GNSS) reflectometry for monitoring lake ice phenology. This is demonstrated using Qinghai Lake, Tibetan Plateau, as a case study. Signal-to-noise ratio (SNR) values obtained from the cyclone GNSS (CYGNSS) constellation over four ice seasons (2018 to 2022) were used to examine the impact of lake surface conditions on reflected GNSS signals during open water and ice cover seasons. A moving t-test algorithm was applied to time-varying SNR values allowing for the detection of lake ice at daily temporal resolution. Good agreement was achieved between ice phenology records derived from CYGNSS data and Moderate Resolution Imaging Spectroradiometer (MODIS) imagery. The CYGNSS timings for freeze-up, i.e., the period starting with the first appearance of ice on the lake (freeze-up start; FUS) until the lake becomes fully ice covered (freeze-up end; FUE), as well as those for breakup, i.e., the period beginning with the first pixel of open water (breakup start; BUS) and ending when the whole lake becomes ice-free (breakup end; BUE), were validated against the phenology dates derived from MODIS images. Mean absolute errors are 7, 5, 10, 4, and 5 days for FUS, FUE, BUS, BUE, and ice cover duration, respectively. Observations revealed the sensitivity of GNSS reflected signals to surface melt prior to the appearance of open water conditions as determined from MODIS, which explains the larger difference of 10 days for BUS.
      PubDate: TUE, 07 NOV 2023 09:17:13 -04
      Issue No: Vol. 17, No. null (2023)
       
  • The Transient Electromagnetic Response of UXO in Complex Three-Dimensional
           Terrain

    • Free pre-print version: Loading...

      Authors: Taoming Lu;Huotao Gao;Shengjie Lv;Wang Yao;Yunkun Yang;Bin Zhang;Yanjie Xiang;
      Pages: 674 - 684
      Abstract: In practical exploration, the complex and diverse topographic effects are important for the processing and interpretation of EM exploration data. The interpretation of the electromagnetic data results in large errors when the influence of topographical factors is simply ignored. At present, the study of TEM is mainly based on the ideal situation of flat terrain, and the study of topographic effect is less. However, topographic relief is inevitable in practical exploration. In this article, the correctness of the time-domain 3-D forward evolution based on the finite-element method (FEM) is verified by comparing the numerical results with those of the homogeneous full-space and half-space analytical solutions and the lumped high conductor model. In this article, 3-D numerical simulations of time-domain transient electromagnetic methods are calculated based on the FEM. Several exploration scenarios for complex terrain were constructed under the transmitting loop source, and the effect of terrain size and filling medium changes on the electromagnetic response was analyzed. The parameters of the topographic effect, earth resistivity, surface resistivity, and target conductivity are analyzed from the z-component of the magnetic field, apparent resistivity, and time-domain numerical solution, and the relative error distribution results in the whole observation period are shown. The results show that the influence of topographic effects on the electromagnetic response is concentrated in the early part of the time domain, with puddles of the same size having a higher influence on the electromagnetic response than raised soils and hollows. In practical exploration, puddles should be avoided where possible, whereas small-sized topographic reliefs and hollows have a negligible effect on the electromagnetic response.
      PubDate: MON, 13 NOV 2023 09:17:21 -04
      Issue No: Vol. 17, No. null (2023)
       
  • A Lightweight Recurrent Aggregation Network for Satellite Video
           Super-Resolution

    • Free pre-print version: Loading...

      Authors: Han Wang;Shengyang Li;Manqi Zhao;
      Pages: 685 - 695
      Abstract: Intelligent processing and analysis of satellite video has become one of the research hotspots in the representation of remote sensing, and satellite video super-resolution (SVSR) is an important research direction, which can improve the image quality of satellite video. However, existing approaches for SVSR often underutilize a notable advantage inherent to satellite video, the presence of extensive sequential imagery capturing a consistent scene. Presently, the majority of SVSR methods merely harness a limited number of adjacent frames for enhancing the resolution of individual frames, thus resulting in suboptimal information utilization. In response, we introduce the recurrent aggregation network for satellite video superresolution (RASVSR). This innovative framework leverages a bidirectional recurrent neural network to propagate extracted features from each frame across the entire video sequence. It relies on an alignment method based on optical flow and deformable convolution (DCN) to realize the alignment of the features, and a temporal feature fusion module to realize effective feature fusion over time. Notably, our research underscores the positive influence of employing lengthier image sequences in SVSR. In the context of RASVSR, with better alignment and fusion, we make the perceptual field of each frame spanning 100 frames of the video, thus, acquiring richer information, and information between different images can be complementary. This strategic approach culminates in superior performance compared with alternative methods, as evidenced by a noteworthy 1.15 dB improvement in PSNR, with very few parameters.
      PubDate: MON, 13 NOV 2023 09:17:21 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Synthetic Aperture Scatter Imaging

    • Free pre-print version: Loading...

      Authors: Qian Huang;Zhipeng Dong;Gregory Nero;Yuzuru Takashima;Timothy J. Schulz;David J. Brady;
      Pages: 696 - 704
      Abstract: Diffraction limits the minimum resolvable feature on remotely observed targets to $\lambda R_{c}/A_{c}$, where $\lambda$ is the operating wavelength, $R_{c}$ is the range to the target and $A_{c}$ is the diameter of the observing aperture. Resolution is often further reduced by scatter or turbulence. Here we show that analysis of scattered coherent illumination can be used to achieve resolution proportional to $\lambda R_{s}/A_{s}$, where $R_{s}$ is the range between the scatterer and the target and $A_{s}$ is the diameter of the observed scatter. Theoretical analysis suggests that this approach can yield resolution up to 1000× better than the diffraction limit. We present laboratory results demonstrating $>30\times$ improvement over direct observation. In field experiments, we use a 23.5 cm aperture telescope at 100 m to resolve 27.78 $\mu$m features, improving on diffraction limited resolution by $>10\times$. The combination of lab and field results demonstrates the potential of scatter analysis to achieve multiple order of magnitude improvements in resolution in applications spanning microscopy and remote sensing.
      PubDate: FRI, 03 NOV 2023 09:16:47 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Separating and Unwrapping of Residual Topographical and Time-Series
           Deformation Phases in PS-InSAR Based on Zero-Temporal Baseline Combination
           and Time-Domain Differentiation

    • Free pre-print version: Loading...

      Authors: Jingxin Hou;Bing Xu;Zhiwei Li;Qi Chen;Yan Zhu;Wenxiang Mao;
      Pages: 705 - 718
      Abstract: To separate and unwrap the residual topographical and deformation components from the wrapped phases are the key issues for persistent scatterer interferometric synthetic aperture radar (PS-InSAR) technique. The current methods, such as grid-search method based on the assumption that the deformation in time-domain is linear, or the least squares ambiguity decorrelation adjustment method that needs priori information to construct the pseudo-observations, etc., are still difficult to tradeoff accuracy against efficiency. In this article, we propose a new method to separate and unwrap the differential phases for PS-InSAR by dividing them into residual topographical phase and differential deformation phases and performing the phase unwrapping separately. First, we calculate phase ambiguities caused by residual topographical error by zero-temporal baseline combination and single parameter grid-search method. The calculated residual topographical error is used to separate the differential phases dominated by deformations from the original differential phases of PS-InSAR. Then, we proposed time-domain differentiation to unwrap differential phases dominated by deformations without any assumption of deformation model in time-domain. The performance of the proposed method is tested by both simulated and real datasets. The experimental results show that, compared with classic PS-InSAR, the accuracy of residual topographical error estimation is comparable; however, there is an accuracy improvement of 34.4% for the deformation parameter estimation and 35.7% for nonlinear deformation time-series extraction, as well as an improvement of 20–40 times in computational efficiency.
      PubDate: TUE, 28 NOV 2023 09:20:57 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Concatenated Deep-Learning Framework for Multitask Change Detection of
           Optical and SAR Images

    • Free pre-print version: Loading...

      Authors: Zhengshun Du;Xinghua Li;Jianhao Miao;Yanyuan Huang;Huanfeng Shen;Liangpei Zhang;
      Pages: 719 - 731
      Abstract: Optical and synthetic aperture radar (SAR) images provide complementary information to each other. However, the heterogeneity of same-ground objects brings a large difficulty to change detection (CD). Correspondingly, transformation-based methods are developed with two independent tasks of image translation and CD. Most methods only utilize deep learning for image translation, and the simple cluster and threshold segmentation leads to poor CD results. Recently, a deep translation-based CD network (DTCDN) was proposed to apply deep learning for image translation and CD to improve the results. However, DTCDN requires the sequential training of the two independent subnetwork structures with a high computational cost. Toward this end, a concatenated deep-learning framework, multitask change detection network (MTCDN), of optical and SAR images is proposed by integrating the CD network into a complete generative adversarial network. This framework contains two generators and discriminators for optical and SAR image domains. Multitask refers to the combination of image identification by discriminators and CD based on an improved UNet++. The generators are responsible for image translation to unify the two images into the same feature domain. In the training and prediction stages, an end-to-end framework is realized to reduce cost. The experimental results on four optical and SAR datasets prove the effectiveness and robustness of the proposed framework over eight baselines.
      PubDate: FRI, 17 NOV 2023 09:19:14 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Two-Stage Evolutionary Algorithm Based on Subspace Specified Searching for
           Hyperspectral Endmember Extraction

    • Free pre-print version: Loading...

      Authors: Cong Lei;Rong Liu;Ye Tian;
      Pages: 732 - 747
      Abstract: In recent years, the introduction of multiobjective evolutionary algorithms (MOEAs) into the field of endmember extraction (EE) in hyperspectral unmixing has demonstrated a breadth of results that surpass those derived from single-objective-based methodologies. Despite these advancements, the adaptation of MOEAs to EE and the attainment of globally optimal solutions represent unresolved challenges meriting continued exploration. This study addresses two principal obstacles in MOEA-based EE: the notorious “curse of dimensionality” in high-dimensional optimization, and the difficulty in striking a balance between convergence and population diversity. We propose a two-stage, evolutionary-based EE algorithm, referred to as TSEA, designed to confront these issues. A novel solution space splitting strategy is incorporated into TSEA that efficiently mitigates the curse of dimensionality by strategically contracting the search space. This advantage is largely attributed to the significant reduction of invalid solutions achieved through the simple application of a clustering procedure. Furthermore, a two-stage optimization approach is employed to meticulously uphold the convergence and diversity of the population, aiming to attain the optimal solution within the realm of high-dimensional optimization. Empirical evidence from four real hyperspectral images demonstrates that the proposed TSEA outperforms other comparison multiobjective optimization algorithms. Thus, this study contributes to the ongoing discourse on the optimization and applicability of MOEAs in the context of EE.
      PubDate: FRI, 17 NOV 2023 09:19:12 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Microwave Radiometer Calibration Using Deep Learning With Reduced
           Reference Information and 2-D Spectral Features

    • Free pre-print version: Loading...

      Authors: Ahmed Manavi Alam;Mehmet Kurum;Mehmet Ogut;Ali C. Gurbuz;
      Pages: 748 - 765
      Abstract: The accuracy of geophysical retrievals from radiometers relies on calibration quality, encompassing both absolute radiometric accuracy and spectral consistency. Radiometers have employed various calibration techniques, including external targets, vicarious sources, and internal calibrators like noise diodes or matched reference loads. Calibration techniques face challenges like frequency dependence, instrumental effects, environmental influences, drift, aging, and radio frequency interference. Recent hardware advancements enable radiometers to collect raw samples containing both temporal and spectral information. Leveraging advanced modeling techniques like deep learning (DL) enables detecting subtle correlations, non-linear dependencies, and higher-order interactions within the data extracting valuable information that may have been challenging with conventional methods. This study utilizes NASA's Soil Moisture Active Passive (SMAP) satellite's level 1A and level 1B data products to develop a DL-based radiometer calibrator to estimate antenna temperature. Spectrograms of second raw moments equivalent to power carrying the 2-D spectral features serve as primary input in a supervised convolutional neural network-based architecture. DL-based calibrator has demonstrated high correlation and low root mean square error when incorporating spectral information from both reference and noise diodes and when not considering this information. Findings suggest that the ancillary features such as internal thermistor temperature and loss elements exhibit sufficient accuracy in estimating antenna temperature to compensate for variations in receiver noise temperature and short-term gain fluctuations in the absence of the reference load and noise diode power. The proposed calibration technique with reduced reference information might enable radiometers for a higher number of antenna scene observations within a footprint.
      PubDate: WED, 15 NOV 2023 09:16:35 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Complex Permittivity Imaging by Incorporating Synthetic Aperture Radar and
           Inverse Scattering Method for Stratified Ground Medium

    • Free pre-print version: Loading...

      Authors: Yoshihiro Yamauchi;Shouhei Kidera;
      Pages: 766 - 777
      Abstract: This article introduces the incorporationapproach with synthetic aperture radar (SAR) and contrast source inversion (CSI) based nonlinear inverse scattering (NIS) approach for quantitative permittivity imaging for buried object under multilayered heterogeneous ground media. It is challenging issue to retrieve a complex permittivity from ground-penetrating radar (GPR) data, since the NIS problem considerably suffers from inaccuracy due to severe ill-posed condition. To overcome this limitation, this article introduces the SAR image-based region of interest (ROI) limitation in the CSI optimization scheme, where the number of unknowns are massively reduced. Furthermore, the SAR image is also upgraded by the pre-CSI optimization, where the Green's function and background clutter for heterogeneous background (e.g., multilayered medium) are accurately generated. The FDTD-based numerical tests, assuming GPR observation model, show that our proposed scheme effectively reconstructs a dielectric property of buried object, even in severe condition.
      PubDate: TUE, 28 NOV 2023 09:20:57 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Feature Enhancement and Alignment for Oriented Object Detection

    • Free pre-print version: Loading...

      Authors: Xu Xie;Zhi-Hui You;Si-Bao Chen;Li-Li Huang;Jin Tang;Bin Luo;
      Pages: 778 - 787
      Abstract: Over the last few years, developments in fields such as aviation and remote sensing have drawn increasing attention to the detection of rotated objects. Unlike general object detection, rotated object detection requires overcoming certain challenges, such as detecting objects with different directions and high aspect ratios. Recently proposed rotated object detectors have achieved good results, but most of them rely on hand-designed anchors, which require manual adjustment of the anchors settings in different scenarios. On the contrary, this article presents a method called FEADet, an anchor-free detector that utilizes feature enhancement and alignment to achieve competitive performance, without the use of anchors. Specifically, in order to better fuse features across different layers, we design an attention feature fusion (AAF) module to reduce the aliasing effect produced by the fusion of different layers. To deal with feature misalignment in detecting objects with orientation, we propose an adaptive alignconv (AAC) module, which is implemented by the constrained deformable convolution and align convolution. The ACC module can efficiently extract object features according to the decoded boxes and predicted constrained offsets. On the two benchmark datasets, dataset for object detection in aerial images (DOTA) and high resolution ship collection 2016 (HRSC2016), a comprehensive evaluation of our method has been conducted to demonstrate the effectiveness of these method in comparison with state-of-the-art methods.
      PubDate: FRI, 17 NOV 2023 09:19:14 -04
      Issue No: Vol. 17, No. null (2023)
       
  • CombineDeepNet: A Deep Network for Multistep Prediction of Near-Surface
           PM$_{2.5}$ Concentration

    • Free pre-print version: Loading...

      Authors: Prasanjit Dey;Soumyabrata Dev;Bianca Schoen Phelan;
      Pages: 788 - 807
      Abstract: PM$_{2.5}$ is a type of air pollutant that can cause respiratory and cardiovascular problems. Precise PM$_{2.5}$ ($\mu {\text {g/m}} ^{3}$) concentration prediction may help reduce health concerns and provide early warnings. To better understand air pollution, a number of approaches have been presented for predicting PM$_{2.5}$ concentrations. Previous research used deep learning models for hourly predictions of air pollutants due to their success in pattern recognition, however, these models were unsuitable for multisite, long-term predictions, particularly in regard to the correlation between pollutants and meteorological data. This article proposes the combine deep network (CombineDeepNet), which combines multiple deep networks, including a bidirectional long short-term memory, bidirectional gated recurrent units, and a shallow model represented by fully connected layers, to create a hybrid forecasting system. It can effectively capture the complex relationships between air pollutants and various influencing factors to predict hourly PM$_{2.5}$ concentrations in multiple monitoring sites based in China. The best root mean square error achieved was 22.0 $\mu {\text {g/m}} ^{3}$ (long-term) and 6.2 $\mu {\text {g/m}} ^{3}$ (short-term), with mean absolute error values of 3.4 $\mu {\text {g/m}} ^{3}$ (long-term) and 2.2 $\mu {\text {g/m}} ^{3}$ (short-term). In addition, the correlation coefficient (R$^{2}$) reached 0.96 (long-term) and 0.83 (short-term) across six monitoring sites. These results demonstrate that CombineDeepNet enhances prediction accuracy compared with popular deep learning methods. Therefore, CombineDeepNet proves to be a important framework for predicting PM$_{2.5}$ concentration.
      PubDate: THU, 16 NOV 2023 09:19:03 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Machine Learning-Based Wind Direction Retrieval From Quad-Polarized
           Gaofen-3 SAR Images

    • Free pre-print version: Loading...

      Authors: Weizeng Shao;Yuhang Zhou;Qingjun Zhang;Xingwei Jiang;
      Pages: 808 - 816
      Abstract: In this article, an intelligent method for inverting wind direction from quad-polarized Gaofen-3 (GF-3) synthetic aperture radar (SAR) images is proposed. Specifically, 11300 acquired in wave (WAV) mode are used to retrieve the wind directions using a spectrum-transformation approach and prior information from European Centre for Medium-Range Weather Forecasts reanalysis at version 5 (ERA-5) data at 1-h intervals with 0.25° grids. The dependence of the wind direction on the polarimetric correlation coefficient (PCC) between the co- (vertical-vertical and horizontal-horizontal) and cross-polarization [vertical-horizontal and horizontal-vertical] channels is studied. It is found that the PCCs in four combination polarizations have asymmetric characteristics with respect to the wind direction with correlation coefficients of greater than 0.4 or less than −0.4. Following this rationale, the scheme for inverting wind direction from quad-polarized SAR is trained according to machine learning, in which the matrix PCCs, wind directions, azimuthal angles, and slopes from SAR intensity spectra at the peaks are used as inputs. Subsequently, this intelligent approach is applied to 1300 images in quad-polarization stripmap mode, and the retrieval results are validated against advanced scatterometer (ASCAT) measurements. The statistical analysis shows that the root-mean- squared error (RMSE) of the wind direction is 17.7°, the COR is 0.98, and the scatter index is 0.11. In addition, the wind speeds inverted using a geophysical model function CSARMOD-GF are compared with well-calibrated ASCAT products, resulting in an RMSE of 1.85 m/s, a COR of 0.78, and an SI of 0.28 for the wind speed. Thus, this article provides an automatic scheme for inverting wind from quad-polarized GF-3 SAR images without any external information.
      PubDate: THU, 16 NOV 2023 09:19:02 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Forward Modeling for the Fine Detection of Geological Abnormal Area in
           Coal Seams Using High-frequency Radio Imaging Method

    • Free pre-print version: Loading...

      Authors: Shun Yang;Yanqing Wu;Peng Lu;Zhifang Liu;
      Pages: 817 - 828
      Abstract: The radio imaging method (RIM) can be used in underground mines to identify geologically hazardous areas; however, it does not yield accurate results. To address this issue, a high-frequency (1–8 MHz) RIM is proposed here for the fine detection of geological anomalies in coal seams, and its feasibility is evaluated. A reliable forward model is established based on the propagation characteristics of electromagnetic waves in lossy media and the functional relation between the electrical parameters of coal and electromagnetic wave frequency. Numerical simulations were performed using a frame antenna at high frequencies; the field strengths of electromagnetic waves in coal seams with and without geological abnormal areas were compared. The difference diagram demonstrates the field strength attenuation “shadow area” formed by the geological abnormal area. The attenuation coefficient, wavelength, and far-field radiation characteristics of the frame antenna were used to discuss if the detected geologically hazardous location is accurate. The feasibility of high-frequency detection is further supported by the electromagnetic wave signal strength in the receiving tunnel.
      PubDate: MON, 20 NOV 2023 09:17:30 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Multispectral UAV-Based Monitoring of Cassytha Filiformis Invasion in
           Xisha Islands

    • Free pre-print version: Loading...

      Authors: Yuhan Xie;Wenjin Wu;Xinwu Li;Jiankang Shi;Tong Yu;Xiaohui Sun;
      Pages: 829 - 841
      Abstract: Currently, numerous studies have reported that the invasion of Cassytha filiformis has affected both above and below ground communities, resulting in difficulties in the growth of original vegetation. Meanwhile, Cassytha filiformis was observed on the Xisha Islands in recent years which brings up the importance of monitoring its invasion to protect the biodiversity of the island. Nonetheless, to effectively monitor Cassytha filiformis at finer regional scales, there is a pressing need for centimeter-level resolution, a level of precision that current satellite sensors find challenging to attain in a consistent manner. Therefore, we adopted a DJI Phantom 4 unmanned aerial vehicle with five multispectral bands and centimeter-level spatial resolution to overcome this problem. An advanced deep learning network is employed to identify the invasion in Xisha Islands for three different time periods. Results show that the area of Cassytha filiformis on Bei Island increased from 211.8 m2 in April 2020 to 458.6 m2 in April 2021, and dropped to 112.8 m2 in July 2021, while that on Ganquan Island changed from 1996.9 m2 in April 2021 (dry season) to 1275.9 m2 in July 2021 (wet season). By incorporating climatic indicators, we further found that Cassytha filiformis in both Bei Island and Ganquan Island favors dry climate and its large area invasion in 2021 was possibly caused by a drought event.
      PubDate: TUE, 07 NOV 2023 09:17:13 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Drifting Ionospheric Scintillation Simulation for L-Band Geosynchronous
           SAR

    • Free pre-print version: Loading...

      Authors: Feixiang Tang;Yifei Ji;Yongsheng Zhang;Zhen Dong;Zhaokai Wang;Qingjun Zhang;Bingji Zhao;Heli Gao;
      Pages: 842 - 854
      Abstract: An L-band geosynchronous synthetic aperture radar (GEO SAR) is susceptible to the ionospheric scintillation induced by ionospheric irregularities due to its lower carrier frequency. In particular, the drifting characteristic of the ionospheric irregularities makes this issue more complicated because the drifting velocity is comparable with the scanning velocity of ionospheric penetration points (IPPs) of GEO SAR. In this article, a drifting ionospheric scintillation simulator (DISS) is devised for producing L-band GEO SAR raw data affected by the drifting ionospheric scintillation phase and amplitude errors. The DISS is divided into a static scintillation transfer function (STF) generator, an IPP drifting offset indexer and a GEO SAR echo generator. Each azimuth echo of a point target is modulated by the complex STF versus the drifting IPP coordinate. Finally, two simulation experiments of the point-target and extended-target scenes are conducted to validate the effectiveness of the DISS. The results indicate that the drifting ionospheric irregularities present absolutely different artifacts and more serious decorrelations in L-band GEO SAR images, compared with the traditional L-band low-Earth-orbit (LEO) SAR.
      PubDate: MON, 06 NOV 2023 09:18:29 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Multiscale Adjacency Matrix CNN: Learning on Multispectral LiDAR Point
           Cloud via Multiscale Local Graph Convolution

    • Free pre-print version: Loading...

      Authors: Jian Yang;Binhan Luo;Ruilin Gan;Ao Wang;Shuo Shi;Lin Du;
      Pages: 855 - 870
      Abstract: Multispectral LiDAR can rapidly acquire 3D and spectral information of objects, providing richer features for point cloud semantic segmentation. Despite the remarkable performance of existing graph neural networks in point cloud segmentation, extracting local features still poses challenges in multispectral LiDAR point cloud scenes due to the uneven distribution of geometric and spectral information. To address the prevailing challenges, cutting-edge research predominantly focuses on extracting multiscale local features, compensating for feature extraction shortcomings. Thus, we propose a multiscale adjacency matrix convolutional neural network (MS-AMCNN) for multispectral LiDAR point cloud segmentation. In the MS-AMCNN, a local adjacency matrix convolution module was first proposed to efficiently leverage the point cloud's topological relationships and perceive local geometric features. Subsequently, a multiscale feature extraction architecture was adopted to fuse local geometric features and utilize a global self-attention module to globally model the semantic features of multiscale. The network effectively captures global and local representative features of the point cloud by harnessing the capabilities of convolutional neural networks in local feature modeling and the self-attention mechanism in global semantic feature learning. Experimental results on the Titan dataset demonstrate that the proposed MS-AMCNN network achieves a promising multispectral LiDAR point cloud segmentation performance with an overall accuracy of 94.39% and a mean intersection over union (MIoU) of 86.57%. Compared with other state-of-the-art methods, such as DGCNN, which achieved an MIoU of 85.43%, and RandLA-net, with an MIoU of 85.20%, the proposed approach achieves optimal performance in segmentation.
      PubDate: TUE, 21 NOV 2023 09:16:54 -04
      Issue No: Vol. 17, No. null (2023)
       
  • GateFormer: Gate Attention UNet With Transformer for Change Detection of
           Remote Sensing Images

    • Free pre-print version: Loading...

      Authors: Li-Li Li;Zhi-Hui You;Si-Bao Chen;Li-Li Huang;Jin Tang;Bin Luo;
      Pages: 871 - 883
      Abstract: Extraction of global context information plays a major role in change detection (CD) of remote sensing (RS) images. However, the majority of methods now depend on convolutional neural networks, which are difficult to obtain complete context information due to the limitation of local convolution operation. This study proposes a novel gate attention U-shaped network with transformer for CD of RS images. GateFormer consists of an encoder with transformer-based Siamese network. First, we propose a gate attention mechanism, which filters the low-level information by guiding high-level features and focuses on activation of relevant knowledge instead of allowing all to pass. In addition, space pooling module in generator extracts more spatial features from pixel level to suppress the generation of noises. Finally, in order to increase the CD accuracy of small-scale ground objects, we design a feature downsampling module to minimize the loss of detailed information and compress more small-scale features in feature downsampling of transformer. The efficiency of our suggested approach has been verified by experiments on three RS CD datasets.
      PubDate: TUE, 21 NOV 2023 09:16:53 -04
      Issue No: Vol. 17, No. null (2023)
       
  • DANet-SMIW: An Improved Model for Island Waterline Segmentation Based on
           DANet

    • Free pre-print version: Loading...

      Authors: Jiawei Xu;Jing Li;Xiaoyu Zhao;Kuifeng Luan;Congqin Yi;Zhenhua Wang;
      Pages: 884 - 893
      Abstract: Segmentation of island waterline contributes to shoreline movement analyzing, environmental monitoring, and integrated coastal zone management. To achieve high efficiency and high accuracy of island waterline segmentation in remote sensing images, we proposed a model for island waterline segmentation based on DANet (DANet-SMIW). In DANet-SMIW model, different indexes (Normalized Difference Water Index and OTSU) were taken as new channels added to input dataset, which enhanced waterline's spectral information. DANet backbone network was improved by dense connection of DenseNet. Loss function, consisting of binary cross entropy loss and Dice loss, was used to resolve the sample imbalance problem, and then, rough results of island waterline segmentation were refined by boundary refinement module (BRM). In total, 2042 island images were taken as experiment dataset, which were cropped from Landsat-8 images and divided into 1634 images for training, 100 images for testing, and 308 images for validation, and DANet-SMIW model was compared against other models, including FCN-32s, DeepLabv3+, PSPNet, Dense-ASPP, PSANet, ICNet, DuNet, and PIDNet. Results demonstrated that DANet-SMIW model achieved the highest values with pixel accuracy and Mean Intersection over Union and possessed higher segmentation efficiency than most other models. Collectively, DANet-SMIW model was an integrated accurate and efficient model for island waterline segmentation in remote sensing images.
      PubDate: MON, 13 NOV 2023 09:17:22 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Active–Passive Remote Sensing Identification of Underground Coal Fire
           Zones With Joint Constraints of Temperature and Surface Deformation Time
           Series

    • Free pre-print version: Loading...

      Authors: Yu Chen;Zhihui Suo;Jie Li;Jun Wei;Fei Cao;Huahai Sun;Huaizhan Li;Yandong Gao;Qian Li;Yinglong Yue;Kaimin Xu;
      Pages: 894 - 915
      Abstract: Coal fire is a geological disaster that causes resource waste and environmental pollution globally. Accurate identification of the spatial location of coal fires is critical for effective coal fire governance. However, existing methods for identifying coal fire zones have problems, such as a high omission and misclassification ratio and insufficient consideration of the temporal variation in temperature. Therefore, this article proposed a temporal temperature anomaly extraction algorithm based on adaptive windows (TTAE-AW) to extract temporal temperature anomaly information. Moreover, the spatial coverage of deformation monitoring points was improved using distributed scatterer interferometric synthetic aperture radar (DS-InSAR), and then a double-threshold two-stage filter method (DTTF) was proposed to accurately identify the spatial location of coal fire zones. The Rujigou mining area in Ningxia (China) was chosen as the region of study. Results showed that the temperature anomalies extracted using the TTAE-AW method are more concentrated in coal fire zones and that the amount in different seasons is more stable. The average accuracy and Kappa coefficient were improved by 15.5% and 0.345, respectively, over those of the conventional method. Compared with the small baselines subset InSAR approach, the DS-InSAR technique has 158% higher spatial coverage for monitoring coal fire zones. Compared with in situ observations of coal fire points, the accuracy and Kappa coefficient of the spatial location of fire zones obtained using the DTTF method were 91% and 0.77, respectively, demonstrating that the proposed method can provide reliable technical support for coal fire monitoring and management.
      PubDate: TUE, 28 NOV 2023 09:20:57 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Assessment of High-Resolution LST Derived From the Synergy of Sentinel-2
           and Sentinel-3 in Agricultural Areas

    • Free pre-print version: Loading...

      Authors: Juan M. Sánchez;Joan M. Galve;Héctor Nieto;Radoslaw Guzinski;
      Pages: 916 - 928
      Abstract: This work explores the potential of obtaining high-resolution thermal infrared (TIR) data provided by the Sentinel-2 (S2) & Sentinel-3 (S3) constellation in a typical semiarid agricultural environment. Maps of land surface temperature (LST) with 10–20 m spatial resolution were obtained from the synergy S2–S3 in the Barrax test site in Spain, for a set of 14 different dates in the summers of 2018–2019. Ground measurements of LST transects covering a variety of croplands and surface conditions were used for a ground validation of the disaggregation approaches. A cross validation of the LST products was also conducted using Landsat-8/TIRS images. Two recent approaches exploiting the linkages between shortwave and thermal data were adapted and tested, with differences in the inputs, the physical-mathematical framework, or the treatment of the LST residuals, and two options for the original 1 km S3 LST data were considered. Despite the large range of temperatures registered (295–330 K), differences with observed values resulted in an average RMSE < 3.0 K and a negligible systematic deviation, showing good results even in small fields ∼1 ha. Results confirm the need for appropriate adjustment techniques of the LST residuals obtained to better capture the low temperature conditions. The systematic overestimations introduced by the use of the operational sea and land surface temperature radiometer L2 LST product, and the limitations associated with certain irrigation management are discussed. Results in this work offer a solution to the lack of high-resolution satellite TIR data, and provide new opportunities for LST applications in agricultural areas.
      PubDate: THU, 23 NOV 2023 09:16:44 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Spatial Heterogeneity and Hierarchy of Metropolitan Area Expansion and
           Land Surface Temperature Evolution: A Twin City Perspective

    • Free pre-print version: Loading...

      Authors: Mengqiu Cui;Liang Zhou;Wenda Wang;Dongqi Sun;Bo Yuan;Wei Wei;
      Pages: 929 - 940
      Abstract: Understanding the response of land surface temperature (LST) to urbanization is important for coping with urban warming. However, few studies have discussed the impact of twin-city integration on LST. Here, we explored the relationship between the twin-city integration and LST over the Xi'an-Xianyang Metropolitan Area (XXMA) using the local contour tree algorithm during 2000–2020. Results indicate that: 1) The urban integration of XXMA is obvious, and the spatial scope of XXMA converges toward the Weihe River between the two cities. Besides, its structure tends to be more complex, which develops from 3 levels and 8 centers to 10 levels and 33 centers from 2000 to 2020. 2) The heat islands of XXMA tend to connect and reach the maximum area of 547.43 km2 in 2015, while the LST evolution in two cities is not synchronized. The LST contour tree of Xi'an increased to 6 levels and 26 centers in 2000–2015 and simplified to 5 levels and 17 centers in 2020, while it increased from 1 to 4 levels in Xianyang in 2000–2020. 3) Synergistic effects exist between urban expansion and LST evolution. The correlation coefficient and bivariate Moran's I between nighttime light (NTL) and LST are greater than 0.35, indicating NTL and LST are correlated and have pleasurable spatial consistency. The high–high agglomeration of NTL and LST increases by 132% in area over the 20-year period, showing a trend toward integration. The study results can provide a scientific basis for thermal environment management and urban climate improvement in XXMA.
      PubDate: WED, 18 OCT 2023 09:16:59 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Graph-Constrained Residual Self-Expressive Subspace Clustering Network for
           Hyperspectral Images

    • Free pre-print version: Loading...

      Authors: Kun Huang;Xin Li;Yingdong Pi;Hao Cheng;Guowei Xu;
      Pages: 941 - 955
      Abstract: Hyperspectral images are widely use due to their rich spectral information. Meanwhile, the difficult acquisition of data labels makes unsupervised classification attracts attention. Subspace clustering as an unsupervised classification method is widely used for hyperspectral image analysis because of its excellent performance and robustness. However, conventional subspace clustering does not consider the nonlinear structure of hyperspectral data, and deep subspace clustering tends to ignore the intrinsic structure of hyperspectral data. To address these problems, we developed a self-expressive learning network, ResSENet, for hyperspectral data; we then proposed the application of ResSENet under graph constraints (GC-ResSENet), considering the intrinsic graph structure of the data. Unlike conventional deep subspace clustering, our model discards the self-expressive layer; self-expressive coefficients between datasets are directly solved by the data using network parameters. Hyperparameters are used in the joint loss to balance the self-expressive properties of the data and the graph constraint terms. We evaluated GC-ResSENet by applying it to four well-known datasets, and our network achieved optimal performance. In addition, because of its abandonment of the self-expressive layer, ResSENet is theoretically capable of clustering with large datasets; thus, we evaluated it using two large datasets.
      PubDate: WED, 15 NOV 2023 09:16:35 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Segmenting Individual Trees From Terrestrial LiDAR Data Using Tree Branch
           Directivity

    • Free pre-print version: Loading...

      Authors: Zekun Yang;Yanjun Su;Wenkai Li;Kai Cheng;Hongcan Guan;Yu Ren;Tianyu Hu;Guangcai Xu;Qinghua Guo;
      Pages: 956 - 969
      Abstract: Over the last decade, a number of techniques for individual tree segmentation has been developed for terrestrial laser scanning data. The superpoint segmentation algorithm based on point cloud has been widely used in individual tree segmentation because of its high efficiency and numerous geometric features. However, this algorithm is generally developed for specific tree species and forest types, limiting its universality and performance for different forest types. To handle this problem, a new method based on the topology of tree branches for individual tree segmentation was proposed. Focusing on the general topological structure of trees, the proposed method iteratively assigns each branch based on its directivity to its upper branch at the superpoint level. The proposed method was tested compared with the original superpoint method and an ecological method in six sample plots with different forest conditions. In such plots, the proposed method achieved anticipated performance with an average accuracy of 40% improvements compared with the other two methods, especially in complicated forest conditions. Experimental results also showed an improved average accuracy of 70% compared with the original superpoint method at the point level. This proposed method can effectively expand the universality of the superpoint method to further advance ecological and forest research.
      PubDate: FRI, 17 NOV 2023 09:19:14 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Independent Component Analysis (ICA) Based Method for Estimating the
           Deformation of Highways in Permafrost Region (HPICA)—A Case Study of
           Maduo Section of Gongyu Highway

    • Free pre-print version: Loading...

      Authors: Xuemin Xing;Jiawang Ge;Wei Peng;Jun Zhu;Bin Liu;Jiancun Shi;Guanfeng Zheng;
      Pages: 970 - 984
      Abstract: Highways built in permafrost regions are susceptible to deformations and instability of the roadbed caused by climatic factors. Long-term deformation monitoring is essential to reveal the freeze-thaw-related deformations. When using Interferometric Synthetic Aperture Radar (InSAR) for permafrost highway monitoring, the majority of different physical phase components are usually considered as equally weighted, and the permafrost deformation-related components are mostly modeled with an empirical mathematical model. This may induce uncertainty and difficulties to remove the atmospheric delay and orbital error, which affects both the accuracy and efficiency of deformation estimation. To address these limitations, we propose an independent component analysis (ICA) based method for estimating the deformation of highways in permafrost regions (HPICA). In HPICA, the Fast ICA is utilized to separate the original InSAR unwrapped phases, and then the extracted freeze-thaw deformation-related components are modeled considering the climatic factors. The simulated experiments show that the spatial ICA can more accurately separate the deformation-related signals from the mixed signals than that of temporal ICA. The Maduo section of Gongyu Highway on the Tibetan Plateau was selected as a study area in the real-data experiment. The results showed the maximum cumulative settlement spanning January 2020 to January 2022 was up to –140.8 mm. A comparative analysis indicated that the modeling accuracy of HPICA is with significant improvement. Besides, HPICA could reveal the boundaries of different permafrost regions according to the nature of permafrost, thus assisting in spatial classification of different types of soil regions.
      PubDate: MON, 04 DEC 2023 09:17:31 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Characterizing Near-Nadir and Low Incidence Ka-Band SAR Backscatter From
           Wet Surfaces and Diverse Land Covers

    • Free pre-print version: Loading...

      Authors: Jessica V. Fayne;Laurence C. Smith;Tien-Hao Liao;Lincoln H Pitcher;Michael Denbina;Albert C. Chen;Marc Simard;Curtis W. Chen;Brent A. Williams;
      Pages: 985 - 1006
      Abstract: The Surface Water and Ocean Topography (SWOT) satellite and AirSWOT airborne instrument are the first imaging radar-altimeters designed with near-nadir and low incidence, 35.75 GHz Ka-band InSAR for mapping terrestrial water storage variability. Remotely sensed surface water extents are crucial for assessing such variability but are confounded by emergent and inundated vegetation along shorelines. However, because SWOT-like measurements are novel, there remains some uncertainty in the ability to detect certain land and water classes. This study examines the likelihood of misclassification between 15 land cover types and develops the Ka-band Phenomenology Scattering (KaPS) model to simulate changes to Ka-band backscatter as a result of changing surface water fraction and surface structure, including vegetation morphology and water surface waves. Using a separability metric, the study finds that water is five times more distinct compared with dry land classes, but has the potential to be confused with littoral zone and wet soil cover types. The KaPS scattering model simulates AirSWOT backscatter for incidence angles 1–27°, identifying the conditions under which open water is likely to be confused with littoral zone and wet soil cover types. KaPS characterization of the sensitivity of near-nadir and low incidence Ka-band SAR to small changes in both wet area fraction and surface structure enables a more nuanced classification of inundation area. These results provide additional confidence in the ability of SWOT to classify water inundation extent and open the door for novel hydrological and ecological applications of future Ka-band SAR missions.
      PubDate: WED, 20 SEP 2023 10:11:14 -04
      Issue No: Vol. 17, No. null (2023)
       
  • IceRegionShip: Optical Remote Sensing Dataset for Ship Detection in
           Ice-Infested Waters

    • Free pre-print version: Loading...

      Authors: Peilin Wang;Bingxin Liu;Ying Li;Peng Chen;Peng Liu;
      Pages: 1007 - 1020
      Abstract: As shipping routes and resource exploration move toward high-latitude oceans, sea ice becomes a major threat to the safety of ship navigation, posing significant challenges to the shipping industry and offshore resource development. Continuous development of satellite remote sensing and deep learning has made large-scale and wide-ranging ship detection (SD) possible, which is of great significance for ship safety. However, existing ship datasets used for deep learning only include ship images in open waters (OW), such as ports and inland rivers. Currently, remote sensing datasets suitable for SD in ice-infested waters (IIW) are lacking. SD in IIW is more difficult than SD in OW because of complex background interference from sea ice. Thus, it is infeasible to directly use the features of ships in OW for SD in IIW. Herein, we propose a remote sensing SD dataset called IceRegionShip, which includes subdatasets IceRegionShip–red, green and blue (RGB) and IceRegionShip–ice region ship index (IRSI). IceRegionShip–IRSI consists of low-resolution images processed with IRSI. IceRegionShip–RGB and IceRegionShip–IRSI contain 11 436 and 9073 ship instances, respectively. IRSI was proposed to address false alarms caused by ice interference. To the best of our knowledge, this is the first dataset designed specifically for SD in IIW. In addition, the dataset was evaluated using several advanced detection algorithms, providing a benchmark for SD in IIW and demonstrating the effectiveness of IRSI for SD in low-resolution optical remote sensing images.
      PubDate: WED, 22 NOV 2023 09:16:50 -04
      Issue No: Vol. 17, No. null (2023)
       
  • SASiamNet: Self-Adaptive Siamese Network for Change Detection of Remote
           Sensing Image

    • Free pre-print version: Loading...

      Authors: Xianxuan Long;Wei Zhuang;Min Xia;Kai Hu;Haifeng Lin;
      Pages: 1021 - 1034
      Abstract: With increasingly rapid development of convolutional neural networks, the field of remote sensing has experienced a significant revitalization. However, understanding and detecting surface changes, which necessitate the identification of high-resolution remote sensing images, remain substantial challenges in achieving precise change detection. Excited deep learning-based change detection techniques often exhibit limitations and lack the necessary precision to detect edge details or other nuanced information in remote sensing images. To address these limitations, we propose a unique semantic segmentation deep learning network, the self-adaptive Siamese network (SASiamNet), specifically devised for enhancing change detection in remote sensing images. The SASiamNet excels in real-time land cover segmentation, adeptly extracting local and global information from images via the backbone residual network. Furthermore, it incorporates a primary feature fusion module to extract and fuse the primary stage feature map, and a high-level information refinement module to refine the resultant feature map. This methodology effectively transmutes low-level semantic information into high-level semantic information, thereby improving the overall detection process. Aimed at empirically testing the effectiveness of the SASiamNet, we utilize two distinct datasets: the public dataset, LEVIR-CD, and a challenging dataset, CDD. The latter is composed of bitemporal images sourced from Google Earth, spanning various regions across China. The experiment results unequivocally demonstrate that our approach outperforms traditional methodologies as well as contemporary state-of-the-art change detection techniques, hence underscoring the efficacy of the SASiamNet in the context of remote sensing image change detection.
      PubDate: MON, 06 NOV 2023 09:18:29 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Beyond Supervised Learning in Remote Sensing: A Systematic Review of Deep
           Learning Approaches

    • Free pre-print version: Loading...

      Authors: Benyamin Hosseiny;Masoud Mahdianpari;Mohammadali Hemati;Ali Radman;Fariba Mohammadimanesh;Jocelyn Chanussot;
      Pages: 1035 - 1052
      Abstract: An increasing availability of remote sensing data in the era of geo big-data makes producing well-represented, reliable training data to be more challenging and requires an excessive amount of human labor. In addition, the rapid increase in graphics processing unit processing power has enabled the development of advanced deep learning algorithms, which achieve impressive results in the field of satellite image processing. However, they require a huge and comprehensive training dataset to avoid overfitting problems and to represent a generalizable model. Thus, moving toward the development of nonsupervised deep learning (NSDL) models in different remote sensing applications is an inevitable need. To provide an initial response to that need, this article performs a comprehensive review and systematic meta-analysis of recently published research articles focusing on the applications of NSDL for remote sensing data processing. In order to identify future research directions and formulate recommendations, we extract trends and highlight interesting approaches from this large body of literature. Consequently, current challenges, prospects, and recommendations are also discussed to uncover the trend. According to the results, there is a sharp increasing trend in the applicability of NSDL methods during these few years particularly, with the advent of new deep architectures, such as adversarial, graph, and transformer models. As a result, this review article discusses different remote sensing data processing applications and challenges that can be addressed using NSDL approaches.
      PubDate: TUE, 19 SEP 2023 10:02:31 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Fusion of Surface Soil Moisture Data for Spatial Downscaling of Daily
           Satellite Precipitation Data

    • Free pre-print version: Loading...

      Authors: Qunming Wang;Ping Ji;Peter M. Atkinson;
      Pages: 1053 - 1065
      Abstract: Remote sensing satellites provide an effective solution for obtaining large-scale precipitation data. However, the spatial resolution of satellite-based precipitation products is often too coarse for hydrological applications at the regional scale. As a solution, spatial downscaling has been increasingly investigated to increase the spatial resolution of satellite-based precipitation. The selection of effective explanatory variables at fine spatial resolution has become a crucial concern in precipitation downscaling. Generally, surface soil moisture (SSM) has a strong physical relation with precipitation (especially at the regional scale), but this relationship has rarely been considered. In this article, we proposed to fuse SSM in precipitation downscaling. Specifically, the 3 km SSM data (i.e., SPL2SMAP_S) were incorporated to downscale the 10 km Integrated Multisatellite Retrievals for Global Precipitation Measurement daily precipitation data to 3 km. Based on case studies in southeastern China, the proposed strategy was compared with the existing scheme fusing digital elevation model or normalized difference vegetation index data as an alternative. The results demonstrated that, compared to the original precipitation product, all downscaling results can provide richer spatial details. The proposed scheme outperformed the other schemes, with a correlation coefficient of 0.63, a root-mean-square error of 15.7 mm, and a mean absolute error of 8.74 mm. Furthermore, the proposed scheme is more sensitive to precipitation events of different intensities. In addition, when the historical precipitation is discontinuous, the advantages of the proposed scheme are more apparent.
      PubDate: TUE, 28 NOV 2023 09:20:57 -04
      Issue No: Vol. 17, No. null (2023)
       
  • An Adjacency-Effect-Based Approach for Accuracy Improvement in Satellite
           Land Surface Temperature Disaggregation

    • Free pre-print version: Loading...

      Authors: Mohammad Karimi Firozjaei;Majid Kiavarz;Seyed Kazem Alavipanah;
      Pages: 1066 - 1083
      Abstract: One of the key parameters that affects the accuracy of land surface temperature (LST) disaggregation is the environmental variables that are fed to the disaggregation model. The aim of this article is to present a new strategy for the disaggregation of LST based on adjacency effects. To do this, a dataset obtained from satellite images and auxiliary information from five European cities was used. First, maps of environmental variables that affect LST were collected. Second, a map of effective environmental variables was produced by calculating and applying the influence of the adjacency effects of each environmental variable based on the proposed weighted inverse distance kernel. Finally, the datasets of environmental variables and effective environmental variables were used separately in the disaggregation process to convert LST at 990 m to disaggregated LST (DLST) at 90 m. The mean RMSEs between LST and DLST obtained without considering the adjacency effects approach for the built-up, agricultural, pasture, forest, and water lands in the cold (warm) season were 0.85 (1.55), 0.72 (1.31), 0.98 (1.63), 0.59 (1.2), and 0.40 (1.12) K, respectively. Taking into account the adjacency effects, the mean RMSE between LST and DLST on built-up, agricultural, pasture, forest, and water lands used in the cold season decreased by 0.35, 0.17, 0.13, 0.09, and 0.03 K, respectively. These values were 0.54, 0.36, 0.33, 0.34, and 0.07 K for the warm season, respectively. The result showed that considering adjacency effects increases the accuracy of LST disaggregation.
      PubDate: THU, 19 OCT 2023 09:16:42 -04
      Issue No: Vol. 17, No. null (2023)
       
  • S1S2-Water: A Global Dataset for Semantic Segmentation of Water Bodies
           From Sentinel- 1 and Sentinel-2 Satellite Images

    • Free pre-print version: Loading...

      Authors: Marc Wieland;Florian Fichtner;Sandro Martinis;Sandro Groth;Christian Krullikowski;Simon Plank;Mahdi Motagh;
      Pages: 1084 - 1099
      Abstract: This study introduces the S1S2-Water dataset—a global reference dataset for training, validation, and testing of convolutional neural networks (CNNs) for semantic segmentation of surface water bodies in publicly available Sentinel-1 and Sentinel-2 satellite images. The dataset consists of 65 triplets of Sentinel-1 and Sentinel-2 images with quality-checked binary water mask. Samples are drawn globally on the basis of the Sentinel-2 tile-grid (100 km × 100 km) under consideration of predominant landcover and availability of water bodies. Each sample is complemented with metadata and digital elevation model (DEM) raster from the Copernicus DEM. On the basis of this dataset, we carry out performance evaluation of CNN architectures to segment surface water bodies from Sentinel-1 and Sentinel-2 images. We specifically evaluate the influence of image bands, elevation features (slope) and data augmentation on the segmentation performance and identify best-performing baseline-models. The model for Sentinel-1 achieves an Intersection over Union (IoU) of 0.845, Precision of 0.932, and Recall of 0.896 on the test data. For Sentinel-2 the best model produces an IoU of 0.965, Precision of 0.989, and Recall of 0.951, respectively. We also evaluate the performance impact when a model is trained on permanent water data and applied to independent test scenes of floods.
      PubDate: FRI, 17 NOV 2023 09:19:14 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Spectral–Spatial Classification With Naive Bayes and Adaptive FFT for
           Improved Classification Accuracy of Hyperspectral Images

    • Free pre-print version: Loading...

      Authors: Arvind Kumar Singh;Renuvenkataswamy Sunkara;Govind R. Kadambi;Vasile Palade;
      Pages: 1100 - 1113
      Abstract: This article presents a postprocessing-based spectral–spatial classification (SSC) approach for hyperspectral (HS) images. The approach effectively overcomes the limitations of traditional pixel-based classifiers by integrating spectral and spatial information to achieve improved classification results. Specifically, the proposed method uses principal component analysis to transform the HS images and the Naive Bayes (NB) classifier to quickly derive spectral-posterior probabilities. Spatial-posterior probabilities are then computed using an adaptive fast Fourier transform (AFFT) and a probabilistic closeness function. These probabilities are then combined to generate a precise SSC map. The proposed approach is available in two distinct styles: the conventional NB–AFFT–SSC method and the proposed iterationwise variable sequencing based NB–AFFT–SSC (IVS–NB–AFFT–SSC) method, which classifies one designated class in each iteration. In addition, two wrapper-based feature selection methods are proposed to obtain a set of principal components (PCs) for each class of the HS image, significantly improving classification accuracy. The approach's efficacy is demonstrated through extensive experimentation on three real HS datasets, including Washington DC Mall, Salinas-A, and Botswana. The generality of the approach has been proven through the use of other well-known machine-learning algorithms, such as support vector machine and K-nearest neighbor, as wrappers in the approach. The results confirm that the proposed approach is highly effective, with the IVS approach helping users concentrate on a particular set of PCs for the class of interest.
      PubDate: WED, 25 OCT 2023 09:16:38 -04
      Issue No: Vol. 17, No. null (2023)
       
  • 3-D Sharpened Cosine Similarity Operation for Hyperspectral Image
           Classification

    • Free pre-print version: Loading...

      Authors: Xin Qiao;Swalpa Kumar Roy;Weimin Huang;
      Pages: 1114 - 1125
      Abstract: Due to the advantage of high spectral resolution, hyperspectral imaging techniques have been extensively used in a variety of fields. Hyperspectral images (HSIs) classification is one of the fundamental tasks and attracts significant research interest. HSIs classification is pivotal as it facilitates precise identification of objects, providing invaluable insights for Earth observation tasks, such as resource management and land cover analysis. In existing studies, convolutional operations have been broadly applied for HSIs classification, especially 3-D convolution, which has shown its effectiveness in extracting spectral–spatial features from the raw HSIs. However, HSIs exhibit the characteristic of high dimensionality and pose challenges in extracting more discriminative features. In order to enhance the capability of capturing discriminative spectral–spatial features, in this article, a novel and effective 3-D sharpened cosine similarity (SCS) operation is proposed, serving as a replacement for conventional 3-D convolutional operation in HSIs classification and enhancing the classification accuracy. The 3-D SCS operation calculates and sharpens the cosine similarity between kernels and HSI input data. Based on the 3-D SCS operation, a 3-D SCS neural network is developed for HSIs classification tasks. To evaluate the effectiveness of 3-D SCS operation, experiments are conducted on three real-world HSIs datasets, including the University of Pavia, the University of Trento, and the University of Houston. Quantitative and qualitative experimental results illustrate that the SCS operation can effectively extract discriminative spectral-spatial features, achieving superior performance over the CNNs under the same model configuration.
      PubDate: TUE, 28 NOV 2023 09:20:57 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Road Extraction From Satellite Images Using Attention-Assisted UNet

    • Free pre-print version: Loading...

      Authors: Arezou Akhtarmanesh;Dariush Abbasi-Moghadam;Alireza Sharifi;Mohsen Hazrati Yadkouri;Aqil Tariq;Linlin Lu;
      Pages: 1126 - 1136
      Abstract: These days, extracting information from remote sensing data has a great impact on various aspects of our lives, such as infrastructure and urban planning, transportation and traffic management, forecasting and tracking natural disasters, searching for mineral resources, monitoring environmental changes, and numerous other fields. One crucial application is extracting accurate road information from aerial images, which has many practical applications ranging from our daily lives to long-term planning for transportation systems to autonomous vehicles. Deep learning models have shown great promise in image-processing tasks, specifically in accurately detecting and extracting roads from aerial images. In this study, various techniques were employed to achieve the desired performance. The model is a UNet assisted with attention blocks in the decoder part and trained with a patched, rotated, and augmented dataset that has been extracted from the DeepGlobe dataset. The preprocessing of the dataset included image and mask patching, rotation, exclusion of background-only images, and excluding images with very little road surface. Both patching and background exclusion in preprocessing as hard attention and attention blocks in the model as soft attention were deployed in order to tackle the inherently biased nature of the dataset. This combination of different techniques empowers the proposed model for superior remote sensing image segmentation performance with an accuracy level of 98.33%. In addition to achieve better performance by the model, another objective is to find the issues that cause the model's performance degradation on some image samples. Therefore, a comprehensive analysis of metrics, with a focus on precision and recall as proper metrics for biased dataset analysis, was conducted to identify potential shortcomings in the model or the dataset, and based on the result, several proposals for future work and further investigations were formulated.
      PubDate: TUE, 28 NOV 2023 09:20:57 -04
      Issue No: Vol. 17, No. null (2023)
       
  • VS-LSDet: A Multiscale Ship Detector for Spaceborne SAR Images Based on
           Visual Saliency and Lightweight CNN

    • Free pre-print version: Loading...

      Authors: Hang Yu;Shihang Yang;Suiping Zhou;Yibo Sun;
      Pages: 1137 - 1154
      Abstract: Recently, deep learning-based methods for synthetic aperture radar (SAR) ship detection have made remarkable advancements. However, most existing methods primarily focus on achieving high detection accuracy by employing complex models, leading to an increase in computational costs. In addition, some methods do not adequately consider the impact of speckle noise interference. To address these challenges, we propose a multiscale ship detector, called visual saliency-lightweight ship detector (VS-LSDet), utilizing visual saliency and lightweight convolutional neural network. First, a visual saliency enhancement module is proposed as a preprocessing step to visually highlight the ships and weaken the impact of speckle noise in the image. Second, a lightweight backbone called ghost-shuffle net (GSNet) is designed. We introduce two types of ghost-shuffle blocks as basic convolution blocks by introducing ghost convolution to reduce the model complexity and channel shuffle operation to enhance the representation ability of the feature map. Then, we propose a multishape dilated convolution block incorporated into GSNet to enlarge its receptive fields, further improving the detector's performance. Finally, a hybrid attention module (HyAM) is proposed, it leverages both spatial and channel information within the feature map. HyAM can emphasize ship-related features while suppressing irrelevant features from the background in the feature map. Experimental results on public SAR ship datasets demonstrate that, compared to other ship detection methods, VS-LSDet achieves higher detection accuracy with lower model complexity. Specifically, on the SSDD dataset, the AP value of VS-LSDet is 97.51%, with 2.53 M parameters and 6.21 GFLOPs.
      PubDate: TUE, 28 NOV 2023 09:20:57 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Using Ray Tracing to Improve Bridge Monitoring With High-Resolution SAR
           Satellite Imagery

    • Free pre-print version: Loading...

      Authors: Zahra Sadeghi;Tim Wright;Andrew Hooper;Sivasakthy Selvakumaran;
      Pages: 1155 - 1166
      Abstract: While satellite persistent scatterer synthetic aperture radar (SAR) interferometry is an effective technique to monitor the health of structures via selection of long-term coherent pixels, detailed interpretation of displacement measurements requires knowledge of which surfaces, the reflection is coming from. Ray tracing algorithms can be used to simulate SAR backscatter for structures, and link observed PS pixels to specific parts of structures. We investigate the reflectivity of three bridges in London for a high-resolution TerraSAR-X dataset, using a ray tracing technique. Artificial reflectors are mounted on one of the bridges. We compare the simulated backscatter with the location of points selected as PS pixels using a stack of 38 TerraSAR-X images. The results confirm that we can predict overall scattering behavior of a bridge using SAR simulation techniques when we have access to a 3-D model of the structure. However, the results of simulation depend on the level of details in the 3-D model and a high-detailed 3-D model including corner reflectors allows the ray tracing technique to perfectly simulate position of the strong scatterers. This approach can help designers increase the SAR reflectivity of a bridge in the design phase of structural bridge assets, or in a retrofit phase, by installing artificial reflectors. We also link the strong scatterers in the reflectivity map to the corresponding scattering surfaces in the structural model that contributed to the signal. This allows the end-users of the InSAR products to better understand which sections of a bridge are moving when a PS pixel indicates displacement.
      PubDate: TUE, 28 NOV 2023 09:20:57 -04
      Issue No: Vol. 17, No. null (2023)
       
  • A New BiRNN-SVA Method for Side Lobe Suppression

    • Free pre-print version: Loading...

      Authors: Shuyi Liu;Yan Jia;Yongqing Liu;Limin Zhai;Xiangkun Zhang;
      Pages: 1167 - 1175
      Abstract: The spatially variant apodization (SVA) algorithm, a classic super-resolution method for synthetic aperture radar (SAR) images, can suppress side lobes while maintain the resolution of the main lobe. To address the problem of residual side lobes or loss of main lobe energy in improved SVA algorithms, the article proposes a new side lobe suppression method combining the bidirectional recurrent neural network (BiRNN) and the SVA algorithm, employing BiRNN to extract the main lobe and side lobe features of radar data to achieve side lobe suppression at any Nyquist sampling rate. The land flight experiment data of the fully polarized microwave scatterometer is used to quantitatively evaluate the side lobe suppression performance and the main lobe energy in order to verify the effectiveness of the BiRNN-SVA method. The experimental results demonstrate that the BiRNN-SVA method can be applied to data at any Nyquist sampling rate and has superior PSLR and ISLR compared to the GSVA algorithm and MSVA algorithm. The image processed with the proposed method retains more fine details and edge features. In comparison to the GSVA algorithm and MSVA algorithm, the image contrast and focus have increased by 31.6% and 3.6%, respectively, and by 4.4% and 1.1%.
      PubDate: WED, 06 DEC 2023 09:16:30 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Incorporating Superpixel Context for Extracting Building From
           High-Resolution Remote Sensing Imagery

    • Free pre-print version: Loading...

      Authors: Fang Fang;Kang Zheng;Shengwen Li;Rui Xu;Qingyi Hao;Yuting Feng;Shunping Zhou;
      Pages: 1176 - 1190
      Abstract: Extracting building from high-resolution (HR) remote sensing imagery (RSI) serves a variety of areas, such as smart city, environment management, and emergency disaster services. Previous building extraction methods primarily focus on pixel-level and superpixel-level features, which do not fully utilize the superpixel-level spatial context, leaving room for performance improvement. To bridge the gap, this study incorporates spatial context of both pixels and superpixels for building extraction of HR RSI. Specifically, the proposed method develops a trainable superpixel segmentation module to segment HR RSI into superpixels by fusing pixel features and pixel-level context. And a superpixel-level context aggregation module is devised to incorporate the multiple-scale spatial context of superpixels to extract buildings. Experiments on public challenging datasets show that our method is superior to the state-of-the-art baselines in accuracy, with better building boundaries and higher integrity. This study explores a new approach for HR RSI building extraction by introducing spatial context of superpixels, and a methodological reference for the HR RSI interpretation tasks.
      PubDate: TUE, 28 NOV 2023 09:20:57 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Underground Natural Gas Microleakage Detection With Hyperspectral Imagery
           Based on Temporal Features and Ensemble Learning

    • Free pre-print version: Loading...

      Authors: Jinbao Jiang;Yingyang Pan;Kangning Li;Xinda Wang;Wenxuan Zhang;
      Pages: 1191 - 1203
      Abstract: Microleakage in underground natural gas storage has serious impacts on the environment or public safety. Recent studies have shown that hyperspectral imagery can detect natural gas microleakage by spectral or spatial features of vegetation indirectly. However, the identification of natural gas microleakage based on hyperspectral imagery still suffers from the following problems: the spectral and spatial features of vegetation change in a complex way with increasing stress time; the effectiveness of ensemble classifiers in recognizing natural gas-stressed vegetation in hyperspectral imagery is unclear; and there is also a lack of studies on the spatial and temporal changes of vegetation stress in natural gas microleakage. Therefore, hyperspectral images of wheat, bean, and grass in different periods were collected. First, the spectral features were filtered using the Relief-F algorithm. The spatial texture features were extracted using the grayscale co-occurrence matrix. The temporal features were extracted using the bi-temporal band ratio. Then, an ensemble classification model fusing spectral, spatial, and temporal features was established. Finally, the natural gas microleakage information was extracted based on the minimum external circle, and the spatial-temporal changes of vegetation stress were analyzed. The results showed that the average stress radii of wheat, bean, and grass were 1.07, 0.83, and 0.86 m, respectively. The mean absolute localization error of natural gas microleakage points was less than 0.4 m. This study provides a theoretical basis and technical support for the future use of satellite hyperspectral detection of microleakage in underground gas storage reservoirs.
      PubDate: WED, 22 NOV 2023 09:16:50 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Multiplatform Bundle Adjustment Method Supported by Object Structural
           Information

    • Free pre-print version: Loading...

      Authors: Jianchen Liu;Wei Guo;
      Pages: 1204 - 1214
      Abstract: The registration and integration of data from different platforms are becoming more and more important for real-scene three-dimensional (3-D) reconstruction. In urban areas, the integration of unmanned aerial vehicle images and terrestrial images can compensate for geometric distortions and texture blurring in models generated from single-platform images. However, it remains a question of how to maintain a high accuracy while accounting for the discrepancies between various platforms. Bundle adjustment is a crucial step in building a detailed 3-D model. However, traditional bundle adjustment is usually applied to a single platform. In the case of multiplatform data with significant differences in resolution, flight height, or viewing angles, it can lead to the issues of instability and low accuracy in solving bundle adjustment problems. This article innovatively proposes a multiplatform bundle adjustment method, which is supported by object structural information. First, the method performs patch-based matching of images from different platforms and obtains cross-platform tie points. Second, refined patches obtain object structural information by calculating the depth values and ground sampling distances of image points. Finally, multiplatform bundle adjustment is conducted using weights calculated for both object and image points based on factors obtained in the second step. The experimental results show that, in general, the proposed method can achieve the accuracy level required for practical applications. Compared with the bundle adjustment method without object structural information, the improvement of accuracy is significant, with an average improvement of 53.38% across the four datasets.
      PubDate: TUE, 05 DEC 2023 09:16:33 -04
      Issue No: Vol. 17, No. null (2023)
       
  • A Multi-Input Channel U-Net Landslide Detection Method Fusing SAR
           Multisource Remote Sensing Data

    • Free pre-print version: Loading...

      Authors: Hesheng Chen;Yi He;Lifeng Zhang;Wang Yang;Yaoxiang Liu;Binghai Gao;Qing Zhang;Jiangang Lu;
      Pages: 1215 - 1232
      Abstract: Accurate and efficient landslide identification is an important basis for landslide disaster prevention and control. Due to the diversity of landslide features, vegetation occlusion, and the complexity of the surrounding surface environment in remote sensing images, deep learning models (such as U-Net) for landslide detection based only on optical remote sensing images will lead to false and missed detection. The detection accuracy is not high, and it is difficult to satisfy the demand. Synthetic aperture radar (SAR) has penetrability, and SAR images are highly sensitive to changes in surface morphology and structure. In this study, a multi-input channel U-Net landslide detection method fusing SAR, optical, and topographic multisource remote sensing data is proposed. First, a multi-input channel U-Net model fusing SAR multisource remote sensing data is constructed, then an attention mechanism is introduced into the multi-input channel U-Net to adjust the spatial weights of the feature maps of the multisource data to emphasize the landslide-related features, and finally, the proposed model is applied to the experimental scene for validation. The experimental results demonstrate that the proposed model combined with SAR multisource remote sensing data improves the perception ability of landslide features, focuses on learning landslide-related features, improves the accuracy of landslide detection, and reduces the rate of false detections and missed detections. Compared with the traditional U-Net landslide detection method based on SAR multisource remote sensing data and the traditional U-Net method that disregards SAR multisource remote sensing data, the proposed method has the best quantitative evaluation indicators. Among them, the proposed method obtained the highest F1 value (66.18%), indicating that fused SAR remote sensing data can provide rich and complementary landslide feature information, simultaneously setting up a multichannel U-Net model to input multisource remote sensing data can effectively process landslide feature information. The proposed method can provide theoretical and technical support for landslide disaster prevention and control.
      PubDate: TUE, 05 DEC 2023 09:16:33 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Shadow Detection and Reconstruction of High-Resolution Remote Sensing
           Images in Mountainous and Hilly Environments

    • Free pre-print version: Loading...

      Authors: Zhenqing Wang;Yi Zhou;Futao Wang;Shixin Wang;Gang Qin;Jinfeng Zhu;
      Pages: 1233 - 1243
      Abstract: The undulating terrain in mountainous and hilly regions results in a greater variety and complexity of shadows. Efficient methods for shadow detection and reconstruction in high-resolution remote sensing images are particularly important in such hilly areas. The accurate detection of shadow masks is a prerequisite for shadow reconstruction. By utilizing the features of high hue and low intensity in shadow areas, an initial spectral ratio is constructed based on the CIELCh color space model. Simple linear iterative clustering is employed to perform superpixel segmentation on the image, and the segmented results are spatially constrained to reconstruct the initial spectral ratio. Afterward, an automatic multilevel global thresholding approach is applied to obtain the shadow mask and eliminate the influence of interfering objects. For shadow reconstruction, the segmented superpixels are treated as the smallest processing units. Similar neighboring objects have similar ambient light intensities. Based on this, we propose a shadow reconstruction method, which compensates shadow superpixels using adjacent nonshadow superpixels and determines compensation weights based on their similarity. Furthermore, the shadow boundaries are dilated to obtain penumbra, and mean filtering is performed to compensate for the illumination in the penumbra. Finally, the proposed method is qualitatively and quantitatively compared with existing shadow detection and reconstruction methods. Experimental results demonstrate that this method can accurately detect shadows in high-resolution remote sensing images in mountainous and hilly environments, and effectively reconstruct the spectral information of shadow areas. This has significant implications for subsequent feature extraction and further analysis in mountainous and hilly regions.
      PubDate: MON, 04 DEC 2023 09:17:31 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Efficient 3-D Near-Field MIMO-SAR Imaging Based on Scanning MIMO Array

    • Free pre-print version: Loading...

      Authors: Ze Hu;Dan Xu;Tao Su;Guanghui Pang;Jinrong Zhong;
      Pages: 1244 - 1256
      Abstract: Multiple-input-multiple-output (MIMO) synthetic aperture radar (SAR) (MIMO-SAR) has numerous potential applications in fields, such as medical diagnosis and security screening. In near-field MIMO-SAR imaging, the virtual array is typically defined as the midpoint of the transceiver array. Based on the virtual array, the multistatic data can be converted to monostatic data for imaging. However, these approaches suffer from low accuracy and cannot reconstruct large scenes, despite their small computational load. In this article, we propose a novel near-field MIMO-SAR imaging algorithm based on range compensation. The convolution between transmitting and receiving elements of the MIMO array defines the positions of virtual array's elements, with the center of the small scene selected as the reference point. The range difference between the transceiver array and the corresponding virtual array to the reference point is calculated. An approximate expression of the range difference is also derived, which can effectively reduce the computational load associated with the range compensation. The conditions for partitioning large scenes are derived by analyzing the range difference between the reference point and other scattered points with respect to the antenna elements. Based on the conditions, the large scenes are divided into blocks. The final image is obtained by a weighted sum of each block scene image. Simulation and experimental results on the designed MIMO radar near-field imaging system demonstrate that the proposed algorithm can effectively reconstruct high-resolution scene images of arbitrary size.
      PubDate: WED, 08 NOV 2023 09:16:34 -04
      Issue No: Vol. 17, No. null (2023)
       
  • MLKNet: Multi-Stage for Remote Sensing Image Spatiotemporal Fusion Network
           Based on a Large Kernel Attention

    • Free pre-print version: Loading...

      Authors: Hao Jiang;Yurong Qian;Guangqi Yang;Hui Liu;
      Pages: 1257 - 1268
      Abstract: Currently, within the realm of deep learning-based spatiotemporal fusion algorithms, those that employ solely convolutional operations are unable to efficiently extract the global image information. In addition, fusion networks that employ a combination of convolution and transformer neglect the 2-D structure of remote sensing images and the role of their channels during training, resulting in an increased computational cost. The current complex fusion methods introduce noise and disregard the correlation between low fractional rate image's time-varying features and high-resolution image's spatial features. To address these issues, we propose TFNet—a temporal feature extraction network that combines normal and deep convolutions to better extract temporal features while reducing computational costs. Second, we suggest utilizing a convolution-based attention module with a large kernel to replace the transformer (LAM), which facilitates adjustment in both spatial and channel dimensions while preserving the image structure. Furthermore, for improved image fusion, we recommend a two-stage fusion module to merge feature images of various scales. This module for fusion integrates features of varying scales and resolutions from various perspectives, thereby preventing noise inclusions and producing favourable fusion outcomes. In addition, we advocate for the utilization of spatiotemporal fusion techniques on other satellites by introducing a new dataset, SW, which is founded on satellite images from Gaofen-1 and moderate-resolution imaging spectroradiometer.
      PubDate: MON, 04 DEC 2023 09:17:31 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Robust Multiscale Spectral–Spatial Regularized Sparse Unmixing for
           Hyperspectral Imagery

    • Free pre-print version: Loading...

      Authors: Ke Wang;Lei Zhong;Jiajun Zheng;Shaoquan Zhang;Fan Li;Chengzhi Deng;Jingjing Cao;Dingli Su;
      Pages: 1269 - 1285
      Abstract: With the aid of endmember spectral libraries, sparse unmixing plays a critical role in interpreting hyperspectral remote sensing data. Integrating spatial clues from hyperspectral data into sparse unmixing frameworks is pivotal for enhancing unmixing capabilities. As such, extracting and harnessing spatial signatures from imagery has emerged as a prevalent tactic to optimize unmixing. In real-world scenarios, hyperspectral images are susceptible to noise, which poses great challenges to the separability of ground objects. As a result, most sparse unmixing models are ill-equipped to handle this issue properly, facing risks of failure. To tackle this challenge, we proposed a sparse unmixing technique with robust multiscale spectral–spatial regularization (RMSR). In the proposed RMSR model, an abundance estimation error reduction regularizer and a spectral–spatial weighted sparse regularizer are consolidated into a unified framework, which excavates the spatial information of the image from multiple perspectives. Specifically, in the first part, the abundance estimation error is defined as the difference between the precomputed abundance maps at the superpixel level and the expected abundances calculated from the original data. Then, the $\ell _{2,1}$ norm is applied to it as a regularization term, which enhances the robustness of the model against image noise and outliers. In the second part, image level spectral weighting coefficients and local spatial weighting terms are leveraged to individually enhance the sparsity of the abundance maps and their piecewise smoothness. The experimental results reveal the algorithm's considerable capabilities in noise immunity and improved unmixing abilities.
      PubDate: TUE, 28 NOV 2023 09:20:57 -04
      Issue No: Vol. 17, No. null (2023)
       
  • AEDNet: An Attention-Based Encoder–Decoder Network for Urban Water
           Extraction From High Spatial Resolution Remote Sensing Images

    • Free pre-print version: Loading...

      Authors: Yanjiao Song;Xiaoping Rui;Junjie Li;
      Pages: 1286 - 1298
      Abstract: Accurate water extraction from urban remote sensing images holds great significance in assisting the formulation of river and lake management policies and ensuring the sustainable development of urban water resources. However, urban high-resolution remote sensing images encompass complex spatial and semantic information, which leads to disparities between the extracted water body features based on local and global information, consequently affecting the accuracy of urban water extraction. To tackle this issue, an attention-based encoder–decoder network was proposed. In this network, the backbone employing atrous convolution (AC) facilitated the acquisition of low-level and high-level features of urban remote sensing images at various scales. Integrated with the attention mechanism, the encoder–decoder structure extracted global features in both the spatial and channel domains. Subsequently, these two types of features were merged to yield the urban water segmentation. Moreover, considering both intersection over union and class weights, a joint loss function (JLF) was introduced to further enhance the accuracy of urban water extraction. Experimental results demonstrated the strong performance of the proposed method on both GID and LoveDA datasets.
      PubDate: FRI, 01 DEC 2023 09:19:27 -04
      Issue No: Vol. 17, No. null (2023)
       
  • SpatioTemporal Inference Network for Precipitation Nowcasting With
           Multimodal Fusion

    • Free pre-print version: Loading...

      Authors: Qizhao Jin;Xinbang Zhang;Xinyu Xiao;Ying Wang;Gaofeng Meng;Shiming Xiang;Chunhong Pan;
      Pages: 1299 - 1314
      Abstract: Precipitation plays a significant role in global water and energy cycles, largely affecting many aspects of human life, such as transportation and agriculture. Recently, meteorologists have tried to predict precipitation with deep learning methods by learning from much historical meteorological data. Under this paradigm, the task of precipitation nowcasting is formulated as a spatiotemporal sequence forecasting problem. However, current studies suffer from two inherent drawbacks of the definition of the problem. First, considering that the weather patterns vary in spatial and temporal dimensions, a spatiotemporally shared kernel is not optimal for capturing features across different regions and seasons. Second, these methods isolate the precipitation from other meteorological elements, such as temperature, humidity, and wind. The disability of cross-model learning prevents the possibility of the promotion of precipitation prediction. Therefore, this article proposes a spatiotemporal inference network (STIN) to produce precipitation prediction from multimodal meteorological data with spatiotemporal specific filters. Specifically, we first design a spatiotemporal-aware convolutional layer (STAConv), in which kernels are generated conditioned on the incoming spatiotemporally features vector. Replacing normal convolution with STAConv enables the extraction of spatiotemporal specific information from the meteorological data. Based on the STAConv, the spatiotemporal-aware convolutional neural network (STACNN) is further proposed, fusing the multimodal information, including temperature, humidity, and wind. Then, an encoder–decoder framework composed of RNN layers is built to extract representative temporal dynamics from multimodal information. To investigate the practicality of the proposed method, we employ STIN to predict the following precipitation intensity. Extensive experiments on three meteorological datasets demonstrate the effectiveness of our model on precipitation nowcasting.
      PubDate: FRI, 13 OCT 2023 09:16:31 -04
      Issue No: Vol. 17, No. null (2023)
       
  • An Automatic Spatial-Temporal Evolution Inversion Method of Mining Goaf
           Based on the Improved Hotspot Analysis and Probability Integral Method

    • Free pre-print version: Loading...

      Authors: Lu Li;Jili Wang;Heng Zhang;Yi Zhang;Yingjie Wang;Yuanzhao Fu;
      Pages: 1315 - 1330
      Abstract: Mining activities may cause severe ground subsidence, endangering surface structures, and farmlands. Therefore, the acquisition of spatial-temporal evolution of the mining goaf is of great significance. Hotspot analysis (HSA) based on the Getis-Ord G$_{i}^*$ statistics has been utilized to identify the areas with a rapid deformation rate. In this article, we propose an improved HSA (IHSA) method for automatic extraction of the surface subsidence caused by the mining goaf. In addition, we design a comprehensive workflow for the automatic spatial-temporal evolution inversion of surface deformation induced by the mining goaf. First, time series interferometric synthetic aperture radar (InSAR) is utilized to generate the surface deformation of the mining area. Then, the IHSA method is used for the automatic identification of the mining goaf. Finally, the total least-squares probability integral method (TLS-PIM) is applied for goaf inversion based on the extracted deformation information. For this study, the Wuxiang is selected as the study area. We have compared the IHSA method with four methods using five indicators, and likewise, we have compared the TLS-PIM method with four methods in terms of their correlation with the InSAR results in the strike and dip directions. The experimental results demonstrate the superiority of our method as a support for the geological hazard investigation and mine safety supervision department.
      PubDate: FRI, 17 NOV 2023 09:19:12 -04
      Issue No: Vol. 17, No. null (2023)
       
  • InSAR Simulation and Speeded-Up Robust Features Algorithm for Terrain
           Relative Navigation in PSRs on the Moon

    • Free pre-print version: Loading...

      Authors: Niutao Liu;Ya-Qiu Jin;
      Pages: 1331 - 1337
      Abstract: Landing in the permanently shadowed regions (PSRs) on the Moon requires high-resolution topographic information and accurate navigation. Owing to low Sun elevation angles, there is no direct solar illumination in PSR, making it difficult to acquire high-resolution optical images for terrain relative navigation (TRN). Synthetic aperture radar (SAR) onboard lunar orbiter can acquire the high-resolution digital elevation model (DEM) of PSR with the interference phases from repeat-passes, or alternatively, from multiantenna observations in a single orbit pass. In this article, SAR images from dual-antenna observations obtained in single orbit passes are simulated with two-scale model and Range-Doppler algorithm for the interference phases based on the DEM data from the lunar orbiter laser altimeter (LOLA). Hence, we generate DEMs of PSR in two prominent lunar south polar craters, Shoemaker and Shackleton. After geometric correction, the influence of radar parallax in DEM data are removed. The generated DEM data are used to illustrate the possibility of TRN in PSR with the image-matching algorithm. The slope angle image of the PSR from the generated DEM is taken as the reference image for navigation, while high-resolution slope angle image from LOLA DEM data is taken as the real-time image from the flyer. The speeded-up robust features algorithm matches the feature points in the reference image and real-time image. The location of the matched points determines the position and motion vector of the flyer. The simulation proves the DEM data from InSAR can provide detailed topographic information and can be used for navigation in regions of permanent shadows.
      PubDate: TUE, 05 DEC 2023 09:16:33 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Assessment of the Impact of Long Integration Time in Geosynchronous SAR
           Imagery of Agricultural Fields by Means of GB-SAR Data

    • Free pre-print version: Loading...

      Authors: Albert Aguasca;Antoni Broquetas;Juan M. Lopez-Sanchez;Xavier Fàbregas;Jordi J. Mallorqui Franquet;Mireia Mas;
      Pages: 1338 - 1347
      Abstract: Geosynchronous synthetic aperture radar (GeoSAR) missions offer the advantage of near-continuous monitoring of specific regions on Earth, making them essential for applications that require continuous information. However, wind-induced motion along the inherent long integration time can result in image defocusing, with potential degradation of retrieved information. This article aims to investigate the impact of GeoSAR long integration time in synthetic aperture radar (SAR) imaging and derived products (time series of backscatter and coherence) required to extract agriculture-relevant soil or crop parameters of interest. The study is based on the extensive HydroSoil data acquisition campaign carried out over barley and corn crops, funded by the European Space Agency. The collected raw data are used to synthesize equivalent apertures with integration times of up to 4 h, similar to those acquired with a GeoSAR. These ultraslow apertures facilitate the assessment of the impact of agricultural scene decorrelation on the generation of images with extended integration times.
      PubDate: FRI, 01 DEC 2023 09:19:27 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Instructional Mask Autoencoder: A Scalable Learner for Hyperspectral Image
           Classification

    • Free pre-print version: Loading...

      Authors: Weili Kong;Baisen Liu;Xiaojun Bi;Jiaming Pei;Zheng Chen;
      Pages: 1348 - 1362
      Abstract: Nowadays, an increasing number of hyperspectral images (HSIs) are becoming available. However, the utilization of unlabeled HSIs is extremely low due to high annotation costs. Thus, it is crucial to figure out how to use these unlabeled HSIs and enhance the classification performance. Fortunately, self-supervised training enables us to acquire latent features from unlabeled HSIs, thereby enhancing network performance via transfer learning. Whereas, most current networks for HSIs are inflexible, it is challenging for them to perform learning and accommodate multimodal HSIs. Therefore, we devise a scalable self-supervised network called instructional mask autoencoder, which can extract general patterns of HSIs by these unannotated data. It primarily consists of a spatial–spectral embedding block and a transformer-based masked autoencoder, which are utilized for projecting input samples into the same latent space and learning higher level semantic information, respectively. Moreover, we utilize a random token called $ins\_{t}oken$ to instruct the model learn components of global information, which are highly correlated with the target pixel in HSI samples. In the fine-tuning stage, we design a learnable aggregation mechanism to put all tokens into full play. The obtained results illustrate that our method exhibits robust generalization performance and accelerates convergence across diverse datasets. In cases of limited samples, we conducted experiments on three structurally distinct HSIs, all of which achieved competitive performance. Compared to state-of-the-art methods, our approach demonstrated respective improvements of 1.97%, 0.44%, and 3.35% on these three datasets.
      PubDate: TUE, 28 NOV 2023 09:20:57 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Context Aware Edge-Enhanced GAN for Remote Sensing Image Super-Resolution

    • Free pre-print version: Loading...

      Authors: Zhihan Ren;Lijun He;Jichuan Lu;
      Pages: 1363 - 1376
      Abstract: Remote sensing images are essential in many fields, such as land cover classification and building extraction. The huge difference between the directly acquired remote sensing images and the actual scene, due to the complex degradation process and hardware limitations, seriously affects the performance achieved by the same classification or segmentation model. Therefore, using super-resolution (SR) algorithms to improve image quality and achieve better results is an effective method. However, current SR methods only focus on the similarity of pixel values between SR and high-resolution (HR) images without considering perceptual similarities, which usually leads to the problem of oversmoothed and blurred edge details. Moreover, there is little attention to human visual habits and machine vision applications for remote sensing images. In this work, we propose the context aware edge-enhanced generative adversarial network (CEEGAN) SR framework to reconstruct visually pleasing images that can be practically applied in actual scenarios. In the generator of CEEGAN, we build an edge feature enhanced module (EFEM) to enhance the edges by combining the edge features with context information. Edge restoration block is designed to fuse multiscale edge features enhanced by EFEM and reconstruct a refined edge map. Furthermore, we designed an edge loss function to constrain the generated SR and HR similarity at the edge domain. Experimental results show that our proposed method can obtain SR images with a better reconstruction performance. Meanwhile, CEEGAN can achieve the best results on classification and semantic segmentation datasets for machine vision applications.
      PubDate: WED, 15 NOV 2023 09:16:35 -04
      Issue No: Vol. 17, No. null (2023)
       
  • CSCNet: A Cross-Scale Coordination Siamese Network for Building Change
           Detection

    • Free pre-print version: Loading...

      Authors: Yiyang Zhao;Xinyang Song;Jinjiang Li;Yepeng Liu;
      Pages: 1377 - 1389
      Abstract: Remote sensing image change detection (CD) has witnessed remarkable performance improvements with the guidance of deep learning models, particularly convolutional neural networks and transformers. Current CD methods heavily rely on multilayered backbone structures, such as ResNet and Unet, for feature extraction. However, these approaches exhibit limitations in coordinating the utilization of local and global features across different scales. In this article, we introduce a novel cross-scale coordinated siamese (CSC) network to effectively integrate multiscale information. We introduce a cross-scale coordination module (CSCM) within the CSC network to coordinate internal features of the local branch with cross-scale information from adjacent branches, while simultaneously attending to both the local and global regions. Furthermore, to comprehensively capture contextual information, we propose a transformer aggregation module as a decoder to harmonize the output features of CSCM. We extensively evaluate our proposed CSC network on three datasets, namely, LEVIR-CD, WHU-CD, and GZ-CD. The results demonstrate that our CSC network outperforms other leading methods significantly in terms of F1-score and intersection over union evaluation metrics.
      PubDate: THU, 30 NOV 2023 09:17:21 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Deep-Learning-Based Semantic Segmentation for Remote Sensing: A
           Bibliometric Literature Review

    • Free pre-print version: Loading...

      Authors: Kazi Rakib Hasan;Anamika Biswas Tuli;Md. Al-Masrur Khan;Seong-Hoon Kee;Md Abdus Samad;Abdullah-Al Nahid;
      Pages: 1390 - 1418
      Abstract: Deep learning (DL) has emerged as a powerful technique for a wide range of computer vision applications. Consequently, DL is also being adopted to process geospatial and remote sensing (RS) images. As these methods are sporadic over different studies, many review papers have also been published to gather the approaches and summarize the existing models in this field. However, a state-of-the-art review paper is still scarce in this field that will present a bibliometric analysis as well as a critical analysis of the recent works. Therefore, this article aims to spur the researchers with a bibliometric analysis to identify the current research trend. As a research sample, in total, 281 related papers were collected from the Web of Science source, and bibliometric analysis was accomplished using VOSviewer software. Among the collection of associated works from the database, 28 papers were selected according to the defined criteria for detailed analysis. Besides this, a few research questions were generated to extract necessary information from the literature for extracting the pros and cons of the selected works. DL techniques were applied in these works and achieved results. Furthermore, the papers were also categorized based on the addressed RS application domain.
      PubDate: MON, 30 OCT 2023 09:18:20 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Sparse Principal Component Analysis and Adaptive Multigraph Learning for
           Hyperspectral Band Selection

    • Free pre-print version: Loading...

      Authors: Wenxian Zhang;Aihong Yuan;Jinglei Tang;Xuelong Li;
      Pages: 1419 - 1433
      Abstract: Band selection (BS) is an effective dimensionality reduction technique for hyperspectral images. Although many relevant methods have been proposed, they often only focus on the bandwise information and the correlation between the bands, and few of them pay attention to the manifold preservation in low-dimensional space, which may lead to the intrinsic structure of the data being damaged. In this article, we propose a novel method called sparse principal component analysis and adaptive multigraph learning (SPCA-AMGL) to address this issue. First, it applies SPCA to the BS task to learn the projection weight matrix, which utilizes the orthogonal constraint to remove redundant bands and uses the $ {L}_{\text{2,1}}$ norm to impose a sparse regularization on the weight matrix to ensure the selection of effective bands. Then, to select the bands with manifold preserving capability, AMGL is proposed to capture the local neighbor structure of data by combining the benefit of multiple graphs, which can not only adaptively learn the graph structure but also obtain the analytical solution of the multigraph coefficients. Finally, an alternate iterative algorithm is designed to optimize the proposed method. Abundant experiments on three hyperspectral datasets prove the reliability and superiority of the proposed method.
      PubDate: TUE, 21 NOV 2023 09:16:53 -04
      Issue No: Vol. 17, No. null (2023)
       
  • A Single-Step Multiclass SVM Based on Quantum Annealing for Remote Sensing
           Data Classification

    • Free pre-print version: Loading...

      Authors: Amer Delilbasic;Bertrand Le Saux;Morris Riedel;Kristel Michielsen;Gabriele Cavallaro;
      Pages: 1434 - 1445
      Abstract: In recent years, the development of quantum annealers has enabled experimental demonstrations and has increased research interest in applications of quantum annealing, such as in quantum machine learning and in particular for the popular quantum support vector machine (SVM). Several versions of the quantum SVM have been proposed, and quantum annealing has been shown to be effective in them. Extensions to multiclass problems have also been made, which consist of an ensemble of multiple binary classifiers. This article proposes a novel quantum SVM formulation for direct multiclass classification based on quantum annealing, called quantum multiclass SVM (QMSVM). The multiclass classification problem is formulated as a single quadratic unconstrained binary optimization problem solved with quantum annealing. The main objective of this article is to evaluate the feasibility, accuracy, and time performance of this approach. Experiments have been performed on the D-Wave Advantage quantum annealer for a classification problem on remote sensing data. Results indicate that, despite the memory demands of the quantum annealer, QMSVM can achieve an accuracy that is comparable to standard SVM methods, such as the one-versus-one (OVO), depending on the dataset (compared to OVO: 0.8663 versus 0.8598 on Toulouse, 0.8123 versus 0.8521 on Potsdam). More importantly, it scales much more efficiently with the number of training examples, resulting in nearly constant time (compared to OVO: 85.72 versus 248.02 s on Toulouse, 58.89 versus 580.17 s on Potsdam). This article shows an approach for bringing together classical and quantum computation, solving practical problems in remote sensing with current hardware.
      PubDate: TUE, 28 NOV 2023 09:20:57 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Detection and Mapping of Cover Crops Using Sentinel-1 SAR Remote Sensing
           Data

    • Free pre-print version: Loading...

      Authors: Sami Najem;Nicolas Baghdadi;Hassan Bazzi;Nathalie Lalande;Laurent Bouchet;
      Pages: 1446 - 1461
      Abstract: Cover crops are intermediary crops planted in between two main cash crops. They play a role in limiting nitrate leaching into groundwater. Currently, there is no database pertaining to cover crops despite their importance. The development period of cover crops is characterized by a dense cloud cover in Europe, which obstructs land surface monitoring using optical remote sensing. This study proposes a cover crops mapping method based on synthetic aperture radar remote sensing data from the Sentinel-1 (S1) constellation, which is unaffected by weather conditions. Our method is based on the dynamics of the S1 backscattering coefficient at the plot level. Using a decision tree, we mapped cover crops. In the decision tree algorithm, filters were added to eliminate other crops that temporally intersect with the cover crop, namely wheat and rapeseed. The proposed decision tree proved effective in detecting existing cover crop plots, as shown by the classification Recall values ranging between 83.5% and 95.0% and the high precision values ranging between 81.5% and 89.2%. Comparison with the Random Forest classifier showed that our proposed method yielded better and more consistent results. The main limitations in the classification approach were weak cover crops and residual vegetation. The results show that the developed approach, based on the S1 time series, is capable of remotely monitoring cover crops, giving managers and decision makers the ability to follow farmers’ work and ascertain if they are applying the recommended agricultural practices that promote sustainable land use and limit the contamination of groundwater.
      PubDate: THU, 30 NOV 2023 09:17:21 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Robust Infrared Small Target Detection Using a Novel Four-Leaf Model

    • Free pre-print version: Loading...

      Authors: Dali Zhou;Xiaodong Wang;
      Pages: 1462 - 1469
      Abstract: Infrared small target detection is widely used in the military field, and robust infrared small target detection has significant significance. Inspired by plants, an infrared small target detection method based on the four-leaf model is proposed. This model has both macro and micro attributes, with macro attributes referred to as the background suppressor (BS) and micro attributes referred to as the texture collector (TC). BS is a four-neighborhood model that can achieve background suppression while reducing the interference of bright background clutter in the target neighborhood to a certain extent. TC can collect texture information of small targets and improve the enhancement effect of small targets. The fusion of TC and BS can effectively suppress background clutter and improve the detection performance of infrared small targets. The experiment is carried out on five real infrared image sequences. The results show that the proposed infrared small target detection method can improve the detection rate and reduce the false alarm rate in the face of infrared images with complex backgrounds. Compared to existing algorithms, the algorithm has high robustness.
      PubDate: THU, 30 NOV 2023 09:17:21 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Operational Neural Networks for Parameter-Efficient Hyperspectral
           Single-Image Super-Resolution

    • Free pre-print version: Loading...

      Authors: Alexander Ulrichsen;Paul Murray;Stephen Marshall;Moncef Gabbouj;Serkan Kiranyaz;Mehmet Yamaç;Nour Aburaed;
      Pages: 1470 - 1484
      Abstract: Hyperspectral imaging is a crucial tool in remote sensing, which captures far more spectral information than standard color images. However, the increase in spectral information comes at the cost of spatial resolution. Super-resolution is a popular technique where the goal is to generate a high-resolution version of a given low-resolution input. The majority of modern super-resolution approaches use convolutional neural networks (CNNs). However, convolution itself is a linear operation and the networks rely on the nonlinear activation functions after each layer to provide the necessary nonlinearity to learn the complex underlying function. This means that CNNs tend to be very deep to achieve the desired results. Recently, self-organized operational neural networks (ONNs) have been proposed that aim to overcome this limitation by replacing the convolutional filters with learnable nonlinear functions through the use of MacLaurin series expansions. This work focuses on extending the convolutional filters of a popular super-resolution model to more powerful operational filters to enhance the model performance on hyperspectral images (HSIs). We also investigate the effects that residual connections and different normalization types have on this type of enhanced network. Despite having fewer parameters than their convolutional network equivalents, our results show that ONNs achieve superior super-resolution performance on small HSI datasets.
      PubDate: WED, 15 NOV 2023 09:16:35 -04
      Issue No: Vol. 17, No. null (2023)
       
  • MUSEnet: High Temporal-Frequency Estimation of Landslide Deformation Field
           Through Joint InSAR and Hydrological Observations Using Deep Learning

    • Free pre-print version: Loading...

      Authors: Aoqing Guo;Qian Sun;Jun Hu;Wanji Zheng;Rong Gui;Yana Yu;
      Pages: 1485 - 1499
      Abstract: The Three Gorges hydropower station in China creates a large reservoir by diverting water from the Yangtze River, increasing the risk of geological disasters, especially massive landslides along the reservoir shoreline. To mitigate these risks, improving geological monitoring and early warning systems is crucial. Interferometric Synthetic Aperture Radar (InSAR) is widely used to monitor reservoir bank landslides. However, its potential in early warning systems is limited due to temporal resolution constraints, preventing timely warnings. To address this, we propose integrating daily hydrological data (precipitation and water level observations) with historical InSAR deformation sequences using our deep learning-based multivariate united state estimation network, “MUSEnet.” This approach generates customized daily landslide deformation products for high-risk areas, greatly enhancing early warning capabilities by providing timely and accurate information on landslide occurrence and magnitude. We validated our method using 161 Sentinel-1 A images of the Xinpu landslide in the Three Gorges Reservoir area. Through statistical analysis, we identified different degrees of influence from rainfall and reservoir water level on the deformation of the Xinpu landslide at various locations. Additionally, we observed distinct lag times between deformation and corresponding rainfall and reservoir water level events. By utilizing deep learning, our method estimates nonlinear states by considering hysteresis and intelligently accounts for the impact of rainfall and reservoir water level, resulting in more accurate estimations compared to traditional models.
      PubDate: FRI, 01 DEC 2023 09:19:27 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Optimization of Multi-Subband Parallel and Signal Reconstruction for
           Remote Sensing Satellite Data Transmission

    • Free pre-print version: Loading...

      Authors: Ze Wang;Fangmin He;Zhong Yang;Yunshuo Zhang;Jin Meng;Yaxing Li;
      Pages: 1500 - 1512
      Abstract: The remote sensing satellite is developing toward high resolution, large data capacity, and fast transmission rate. The ground receiver is correspondingly required to have the high parallel to improve real-time reception and processing capabilities. In the frequency domain, the parallel processing of multiple subbands is achieved by dividing broadband into narrow bands. However, narrow-band subbands are reconstructed into broadband signals, which can lead to cross-subband distortion. It will affect the accuracy of high-resolution remote sensing images. In this article, the subband division and reconstruction framework is proposed by combining the analog filter with the digital filter. The phase calibration methods and digital filter optimization are proposed to improve the amplitude and phase consistency of the reconstructed signal. The simulation results show that the proposed amplitude consistency optimization method effectively reduces the reconstruction error within 0.001 dB. The proposed phase calibration method effectively reduces the bit error rate of the reconstructed signal. The maximum deviation is no more than 0.1%. Experiments have shown that the optimization method can reduce the distortion error of high-resolution remote sensing images.
      PubDate: MON, 18 DEC 2023 09:17:56 -04
      Issue No: Vol. 17, No. null (2023)
       
  • HOFA-Net: A High-Order Feature Association Network for Dense Object
           Detection in Remote Sensing

    • Free pre-print version: Loading...

      Authors: YunPeng Xu;Xin Wu;Li Wang;Lianming Xu;Zhengyu Shao;Aiguo Fei;
      Pages: 1513 - 1522
      Abstract: In the remote sensing field, deep learning-based methods have become mainstream for remote sensing image object detection in recent years. However, traditional methods, such as convolutional neural networks (CNNs), mainly ignore the dependencies between features, failing to capture the spatial relationships and relative positions of objects, which affects the detection performance of dense objects, especially small-size objects. To this end, a high-order feature association network (HOFA-Net) for dense object detection in remote sensing has been proposed to better capture the interdependencies between features of channel and spatial dimensions, yielding more distinguishable features. First, we employ CNNs to learn high-level but low-resolution features of objects. To capture feature interdependencies while retaining crucial information, we design a feature association module based on size adaptation nonlocal. This module partitions the low-resolution and high-level features into local regions and utilizes nonlocal residual connections to capture the local contextual information of objects. In addition, we introduce a high-order feature association (HFA) module designed to learn nonlinear feature correlations and interdependencies within the features. In addition, a covariance normalization acceleration strategy is introduced to accelerate computation. Experimental results on two public remote sensing datasets, including the DOTA dataset and the Tiny Person dataset, demonstrate the superiority and effectiveness of the proposed method through comparative experiments.
      PubDate: TUE, 28 NOV 2023 09:20:57 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Multiorder Graph Convolutional Network With Channel Attention for
           Hyperspectral Change Detection

    • Free pre-print version: Loading...

      Authors: Yuxiang Zhang;Rui Miao;Yanni Dong;Bo Du;
      Pages: 1523 - 1534
      Abstract: Hyperspectral change detection (CD) aims to obtain the change information of objects in the multitemporal hyperspectral images (HSIs). Recently, with the advantages in fully extracting the image features of irregular areas, the graph convolutional network (GCN) has attracted increasing attention for hyperspectral CD. The existing GCN-based CD methods usually use a graph structure constructed by superpixels to reduce the computational cost, which ignores the multiorder difference information among graph nodes and the local difference information within superpixels. To address these problems, this article proposes an efficient multiorder GCN with a channel attention module (CAM) for hyperspectral CD. Specifically, the multiorder GCN module is designed by repeatedly mixing the feature representations of neighborhoods. The CAM is then proposed to enhance the difference features of bitemporal HSIs. After that, the pixel-wise CD is accomplished by a lightweight feature fusion module and a fully connected layer. Experiments on three hyperspectral datasets illustrated the effectiveness of the proposed algorithm.
      PubDate: TUE, 05 DEC 2023 09:16:33 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Parallel Fusion Neural Network Considering Local and Global Semantic
           Information for Citrus Tree Canopy Segmentation

    • Free pre-print version: Loading...

      Authors: Haiqing He;Fuyang Zhou;Yuanping Xia;Min Chen;Ting Chen;
      Pages: 1535 - 1549
      Abstract: Existing convolutional neural network (CNN) based methods usually tend to ignore the contextual information for citrus tree canopy segmentation. Although popular transformer models are helpful in extracting global semantic information, they ignore the edge details between citrus tree canopies and the background. To address these issues, we propose a parallel fusion neural network considering both local and global semantic information for citrus tree canopy segmentation from 3-D data, which are derived by unmanned aerial vehicle (UAV) mapping. In the feature extraction stage, a parallel architecture, concatenated by EfficientNet-V2 and CSwin transformer, is used to extract local and global information of citrus trees. In the feature fusion stage, we design a coordinate attention-based fusion module to retain the contextual information and local edge details of citrus tree canopies. Additionally, to exaggerate the exclusivity between tree canopies and complex backgrounds, 3-D data incorporating RGB imagery and canopy height model derived by UAV photogrammetry are generated for citrus tree canopy segmentation. Experimental results indicate that the proposed method performs considerably better than methods based only on CNN or transformer models and is superior to state-of-the-art methods (e.g., the highest mIoU score of 93.46%).
      PubDate: TUE, 05 DEC 2023 09:16:33 -04
      Issue No: Vol. 17, No. null (2023)
       
  • RUW-Net: A Dual Codec Network for Road Extraction From Remote Sensing
           Images

    • Free pre-print version: Loading...

      Authors: Jingyu Yang;Zongliang Gu;Ting Wu;Yousef Ameen Esmail Ahmed;
      Pages: 1550 - 1564
      Abstract: Road information plays an increasingly important role in applications, such as map updating, urban planning, and intelligent supervision. However, roads in remote sensing images may be shaded by trees and buildings or interfered with by farmland. These intrinsic image features can cause road extraction results to suffer from breakage and misidentification problems. To address these problems, this article improves on D-LinkNet and proposes a dual codec structure network, namely RUW-Net. Specifically, we use ReSidual U-blocks instead of ordinary residual blocks to extract more global contextual information during the encoding stage. Moreover, we propose a decoder-encoder combination (DEC) module to build a dual codec structure. The DEC module links the decoder of the first U-block and the encoder of the following U-block to narrow the semantic gap in the encoding and decoding process. The RUW-Net model can extract more multiscale contextual features and effectively use them to enhance the semantic information of road entities. Therefore, the RUW-Net model can obtain more accurate extraction results. We conducted a series of experiments on public datasets, such as DeepGlobe, including comparative, robustness, and ablation experiments. The results show that the proposed model alleviates the road extraction breakage and misidentification problems. Compared with other representative methods, the RUW-Net performs better in terms of completeness and accuracy of road extraction results; overall, its extraction results are also the best. The RUW-Net model provides a new idea for road extraction from remote sensing images.
      PubDate: TUE, 05 DEC 2023 09:16:33 -04
      Issue No: Vol. 17, No. null (2023)
       
  • The ƒ'/Q Factor in the Relationship Between Inherent and Apparent Optical
           Properties in a Typical Inland Water (Lake Taihu, China)

    • Free pre-print version: Loading...

      Authors: Yu Zhang;Lifu Zhang;Yi Cen;Hongying Zhao;Junsheng Li;Qingxi Tong;
      Pages: 1565 - 1574
      Abstract: Water reflectance anisotropy has serious implications for the remote sensing of water surfaces. The ${\bm{f}^{\prime}}/{\bm{Q}}$ factor serves as a crucial bridge connecting the inherent and apparent optical properties of turbid water bodies; ${\bm{f}^{\prime}}/{\bm{Q}}$ varies not only with angular geometry and inherent optical properties (IOPs), but also with wavelength. However, there was limited research on ${\bm{f}^{\prime}}/{\bm{Q}}$ in turbid water bodies. This research conducted preliminary exploration on ${\bm{f}^{\prime}}/{\bm{Q}}$ in inland turbid water bodies based on measurements of in-situ multi-angle bidirectionality reflectance, including 17 angles and 251 bands when the solar zenith angle (${{\bm{\theta }}}_{\bm{s}}$) was between 40° and 50°. This study indicated that ${\bm{f}^{\prime}}/{\bm{Q}}$ exhibited highly sensitivity in the wavelength range of 690–750 nm with a peak near 727 nm, as well as angular geometry played an important role in the BRDF. The ${\bm{f}^{\prime}}/{\bm{Q}}$ varied over the range of 0.13–0.18 sr−1, variation of 0.06 sr−1 across different bands, when viewing zenith angle (${{\bm{\theta }}}_{\bm{v}}$) and particle backscattering ratio (${{\tilde{\bm{b}}}}_{{\bm{bp}}}$) were 0 and 0.0183, respectively; higher than the range observed for the ocean (0.08–0.15 sr−1). The variability in ${\bm{f}^{\prime}}/{\bm{Q}}$ as a function of wavelength must be accounted for in turbid waters. The ${\bm{f}^{\prime}}/{\bm{Q}}$ was correlated with the IOPs in the range of 690–750 nm, with coefficient of determination (R2) values higher than 0.94 and root mean square error (RMSE) values lower than 0.004. Our findings are of great significance for understanding the relationship between inherent and apparent optical properties in inland water bodies, for BRDF studies, and for improving the accuracy of satellite product retrieval.
      PubDate: MON, 18 DEC 2023 09:17:56 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Construction of a High-Quality Digital Elevation Model of the Amundsen
           Crater and Landing Area Selection for Future Lunar Missions

    • Free pre-print version: Loading...

      Authors: Yingjun Zheng;Weifeng Hao;Mao Ye;Yihao Chen;Wensong Zhang;Fei Li;
      Pages: 1575 - 1583
      Abstract: The Amundsen crater is located in the Aitken Basin of the lunar nearside. Its unique location and the possibility of water ice make it a prime landing area for lunar exploration missions. Constructing a high-quality digital elevation model (DEM) and performing a detailed landing site analysis are critical for research and practical applications. However, this region has many permanently shadowed areas, making optical remote sensing observations impractical. The lunar orbiter laser altimeter (LOLA) has provided the highest quality and largest satellite altimetry dataset, making it ideal for constructing a high-quality DEM. High-resolution DEMs derived from LOLA data contain significant noise due to the geographic uncertainty of laser altimetry data. This article utilizes an adaptive iteration method and a filtering method for slope detrending to construct a high-quality DEM of the Amundsen area. Various factors are comprehensively analyzed, including slope, illumination conditions, and temperature. The optimal landing location in the Amundsen area is identified (90.716°E, 84.727°S). The illumination conditions during the landing of the lunar exploration mission are estimated by calculating the obstruction angle of the optimal landing site and the solar altitude angle from 2023 to 2028. The optimal landing time is in October, providing favorable illumination conditions in the coming months. We calculate the maximum range of the azimuth angles (0°–59.5° and 287.5°–360°) that can receive sufficient sunlight at the designated landing site. Our study provides a novel strategy for selecting the placement of solar panels for lunar exploration instruments.
      PubDate: WED, 06 DEC 2023 09:16:29 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Automatic Monitoring of Oil Tank 3D Geometry and Storage Changes With
           Interferometric Coherence and SAR Intensity Information

    • Free pre-print version: Loading...

      Authors: Ya-Lun S. Tsai;Chun-Jia Huang;Chia-Ling Chen;Jen-Yu Han;
      Pages: 1584 - 1595
      Abstract: Continuous monitoring of oil tanks is vital for analyzing local fuel consumption. Synthetic aperture radar (SAR) has been a popular data source as it guarantees day-and-night and all-weather sensing capacity. However, most earlier studies adopt a scene-wise and oil tank-wise scheme, which is inefficient as there can be hundreds of oil tanks on an oil depot, while only a few are dynamic. Also, no study explores both intensity coherence and interferometric coherence for oil tank dynamics mapping. This article proposes a novel three-stage strategy to detect all oil tanks, identify dynamic oil tanks, and estimate their fuel volume changes based on both the intensity and phase information of SAR in both slant-range and geocoded projections. Results indicate that the intensity coherence can perfectly differentiate dynamic and stable oil tanks (a Jeffries–Matusita distance of 1.997) and is less vulnerable to repeat-pass SAR factors, such as baselines and atmospheric conditions. Via evaluating estimations’ consistency, our scattering keypoint detection exhibits 0.23 and 0.87 m precision of tank heights and diameters, respectively. By validation with ground truth data, oil tanks exhibiting floating-roof changes larger than 0.23 m are correctly identified. Also, the estimated storage changes agree well with actual changes with an R-squared value of 0.98 and a root-mean-square error corresponding to 1.05 m biases in floating-roof heights. These quantitative assessments confirm the robustness and broad applicability of our non-in situ data-needed approach, highlighting the opportunity to utilize spotlight SAR data to automatically and comprehensively monitor oil tank dynamics in remote sites.
      PubDate: TUE, 28 NOV 2023 09:20:57 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Near Real-Time Monitoring of Muddy Intertidal Zones Based on
           Spatiotemporal Fusion of Optical Satellites Data

    • Free pre-print version: Loading...

      Authors: Yan Gu;Jianchun Chen;Ziyao Chen;Mingliang Li;Shibing Zhu;Ya Ping Wang;
      Pages: 1596 - 1609
      Abstract: The investigation of the evolution of intertidal zones is a significant and extensively discussed subject in estuary and coastal research, both nationally and internationally. Recent advances in real-time global satellite products, such as GOCI, provide a potential approach for monitoring intertidal zones by utilizing waterline inversion based on spatial distribution processes. While these products offer the potential for monitoring intertidal zones through waterline inversion based on spatial distribution processes, they often suffer from coarse spatial resolution that smooths critical spatial heterogeneity. To overcome this limitation, a flexible spatiotemporal fusion model was employed to generate hourly time-series images with a spatial resolution of 10 m. This was achieved by combining Sentinel-2 satellite data (10 m spatial resolution; five-day revisit frequency) and GOCI-II data (500 m spatial resolution; 1-h revisit frequency). The resulting fusion images were used to construct an intertidal pseudodigital elevation model (DEM) by extracting waterlines and incorporating tidal-level information. The accuracy of the DEM was validated using surveyed real-time kinematic and drone data, with a root-mean-square error of 0.28 m. The analysis of the annual accretion and erosion evolution in the intertidal zone for the year 2023 revealed significant erosion in the central part of the zone, with a maximum erosion depth of 1 m at the bottom. This study contributes to the understanding of the response processes and mechanisms of the intertidal zone to natural and human disturbances, thus supporting coastal planning projects related to the intertidal zone.
      PubDate: TUE, 05 DEC 2023 09:16:33 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Ecological Impact Patterns and Temporal Cycles of Green Tide Biomass in
           the Settlement Region: Based on Time-Series Remote Sensing and In Situ
           Data

    • Free pre-print version: Loading...

      Authors: Guangzong Zhang;Lifeng Niu;Mengquan Wu;Hermann Kaufmann;Hanyu Li;Yufang He;Bo Chen;
      Pages: 1610 - 1622
      Abstract: Recurring green tides (also called Ulva prolifera) cause significant damage to marine ecosystems in the Yellow Sea of China, especially in the settlement region. The settlement region is a critical area for the natural decay and decomposition of green tides, which obviously has ecological consequences. Recent studies in relation to this topic are mostly based on point observation data, which prevents to quantitatively analyze the ecological impact patterns of green tide biomass at large spatial scales. Therefore, we used remote sensing time-series of Geostationary Ocean Color Imager satellite data combined with in situ data to invert marine chlorophyll-a (Chl-a, main indicators representing phytoplankton biomass) concentrations by machine learning methods. Finally, based on a cross-satellite model, we quantified the green tide biomass that occurred in Haiyang City during 2015 and 2016. The main results found are as follows. First, the green tide biomass reveals negative correlations on Chl-a concentration in the settlement region. The Chl-a concentration showed a decreasing trend and remained at a low level (0.1–0.2 mg/m3) when the biomass of the green tide increased. After the disappearance of the green tides, the Chl-a concentration accreted rapidly and then began to gradually decrease. Second, combined with the pixel-based statistics grid of the green tide and the average Chl-a concentration, the time cycle of green tides in the settlement region is about 30–35 d. Finally, some special cases (such as typhoon) can change the pattern and temporal cycle in the settlement region. This article provides support for marine ecosystem monitoring.
      PubDate: MON, 04 DEC 2023 09:17:31 -04
      Issue No: Vol. 17, No. null (2023)
       
  • MSRSI-TPMF: A Tie Points Matching Framework of Multisource Remote Sensing
           Images

    • Free pre-print version: Loading...

      Authors: Qian Cheng;Xin Li;Taoyang Wang;Boyang Jiang;He Fu;Yunming Wang;Feida Zhang;
      Pages: 1623 - 1637
      Abstract: Remote sensing sensor platforms are typically located at a significant distance from the ground, ranging from several hundred meters to hundreds of kilometers. This means that, compared to natural images, remote sensing images (RSI) have larger coverage areas and more complex information. The larger size and data volume of RSI presents challenges for computer vision matching algorithms (MAs), making it difficult to apply them directly to RSI matching. Moreover, a matching framework for multisource RSIs capable of large-scale processing by integrating multiple MAs with the entire RSI as input is presently lacking. This study proposes a tie points (TPs) matching framework of multisource remote sensing images based on the geometric and radiation characteristics of RSI. First, RSI is divided into different grids and undergoes local geometry correction. Next, matching between slice images is performed by MAs. Finally, TPs are generated by mapping matched points in multiple slice images to the whole RSI using a geometric processing model. Six representative MAs including artificial feature MAs and deep learning algorithms are integrated into the framework to match TPs from different RSI. Results demonstrate the extraction of TPs for multisource RSI, validating the framework's efficacy. In addition, a large-scale TPs matching test for deep learning MA is performed by using 13 synthetic aperture radar images (10-m resolution) with TPs root mean square error of 0.368 pixels, further confirming the framework's reliability.
      PubDate: WED, 08 NOV 2023 09:16:34 -04
      Issue No: Vol. 17, No. null (2023)
       
  • A Label Correction Learning Framework for Gully Erosion Extraction Using
           High-Resolution Remote Sensing Images and Noisy Labels

    • Free pre-print version: Loading...

      Authors: Chunhui Zhao;Yi Shen;Nan Su;Yiming Yan;Shou Feng;Wei Xiang;Yong Liu;Tianhao Zhao;
      Pages: 1638 - 1655
      Abstract: Rapid and reliable gully erosion (GE) extraction from high-resolution remote sensing (HRRS) images is crucial for the development of land protection measures. For this task, semantic segmentation methods are widely considered the state-of-the-art solutions. Nevertheless, providing sufficient and clean training labels for segmentation models demands substantial expenses and time investment. In this context, a label correction learning (LCL) framework is proposed to effectively extract GE from HRRS images by leveraging more readily available noisy labels. The core objective of this framework is to suppress the adverse impact of noise within noisy labels on the model performance. To achieve this, we introduce three key components in the framework, including an adaptive correction loss function, a multitree refinement module, and a noise correction module. These components collaborate to rectify noisy labels during training, thereby providing the model with a training set containing less noise. To validate the effectiveness of the LCL framework, three severely eroded regions in Northeast China are selected as study areas and corresponding noisy datasets are generated. Extensive experiments on these datasets demonstrate that our framework can significantly mitigate the negative influence of label noise and ultimately achieve superior GE extraction performance. Moreover, by employing the proposed framework, we generate GE coverage maps for the study areas and obtain measurements of gully area and length that are very close to the true statistics. Such a framework that can effectively learn from noisy labels holds promise as a practical and cost-efficient means to provide reliable data references for land resource protection.
      PubDate: MON, 04 DEC 2023 09:17:31 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Reactivation/Acceleration of Landslides Caused by the 2018 Baige
           Landslide-Dammed Lake and its Breach Floods Revealed by InSAR and Optical
           Images

    • Free pre-print version: Loading...

      Authors: Zhiqiang Xiong;Guangcai Feng;Lijia He;Haiyan Wang;Jianchao Wei;
      Pages: 1656 - 1672
      Abstract: Landslide-dammed lakes and their breaches can reactivate/accelerate landslides, causing potential damages. However, in the absence of displacement observations, the spatial-temporal deformation patterns of the landslide-dammed lakes and associated floods reactivated/accelerated landslides remain underexplored. In this article, we use the 2018 Baige landslide-dammed lake and associated floods reactivated/accelerated landslides to investigate the behaviors of such landslides. We first use an improved interferograms Stacking method to detect landslides, then utilize the multitemporal interferometric synthetic aperture radar to derive their deformation history. Through retrospective analysis of the deformation history and optical images, we find that the water level fluctuations and floods caused by the Baige landslide-dammed lake and its breaches reactivated/accelerated six landslides upstream and five landslides downstream. The area of the largest reactivated/accelerated landslide is about 5 km2. The maximum velocity change of these reactivated/accelerated landslides is about 20 cm/year. Landslide reactivation/acceleration occurs progressively from the toe to head, resulting in varying reactivation/acceleration times for different parts. The velocity change and acceleration area have a linear relationship, with larger landslides showing larger velocity changes and prolonged activity than smaller ones. Among these 11 reactivated/accelerated landslides, 4 are located in the river narrow sections and their volumes are all larger than that of the Baige landslide. Thus, their failure may cause larger damages than that caused by the Baige landslide. Our findings contribute to a better understanding of landslide-induced geological hazard chains and landslide behaviors.
      PubDate: TUE, 05 DEC 2023 09:16:33 -04
      Issue No: Vol. 17, No. null (2023)
       
  • Characterization of the Amazon Rainforest Backscatter at X-Band Using
           TanDEM-X Data

    • Free pre-print version: Loading...

      Authors: Luca Dell'Amore;José–Luis Bueso–Bello;Patrick Klenk;Jens Reimann;Paola Rizzoli;
      Pages: 1673 - 1690
      Abstract: The radiometric calibration of spaceborne SAR products plays a key role for ensuring a good performance of the whole end-to-end system and requires a precise knowledge of both the radar system and the illuminated target on ground. The shape of the antenna pattern in elevation can be directly estimated by analyzing SAR detected images in presence of a h