JournalTOCs

Optical Memory and Neural Networks
Journal Prestige (SJR): 0.276

Citation Impact (citeScore): 1
Number of Followers: 2

Hybrid journal (It can contain Open Access articles)
ISSN (Print) 1060-992X - ISSN (Online) 1934-7898
Published by Springer-Verlag

[2467 journals]

Optical Assessment of In-Office Teeth Whitening Influence on Tooth Enamel
Spectral Composition In Vivo
- Free pre-print version: Loading...
  Rate this result: What is this?
  
  Please help us test our new pre-print finding feature by giving the pre-print link a rating.
  A 5 star rating indicates the linked pre-print has the exact same content as the published article.
  
  Abstract: The results of the in vivo research of tooth enamel using the Raman spectroscopy method are presented in the work. The subject of the study was tooth enamel of patients aged 22–35. The study was carried out before, right after, two weeks and one month after the in-office teeth whitening procedure. The algorithm of assessment of enamel composition after the procedure of in-office teeth whitening in vivo experiments was developed as a result of the study. It was found that teeth enamel spectral changes were insignificant at different time after the in-office teeth whitening, which in turn did not cause the further structural changes of tooth enamel with time. The accuracy and specificity of the developed algorithm were correspondingly 90 and 89%.
  PubDate: 2022-12-01
Patch-Wise Partial Face Recognition Using Convolutional Neural Network
- Free pre-print version: Loading...
  Rate this result: What is this?
  
  Please help us test our new pre-print finding feature by giving the pre-print link a rating.
  A 5 star rating indicates the linked pre-print has the exact same content as the published article.
  
  Abstract: Automatic face recognition still suffers from some problems in the real-world scenarios such as occlusion. Hence, identifying the face from its partial appearance is a challenging issue as yet. To address this, issue many methods have been proposed using traditional feature extraction techniques. In this paper, a partial face recognition problem has been tackled through utilizing patch-wise matching with Convolutional Neural Network (CNN). Firstly, a gallery images are divided into local patches, and each patch is regarded as an independent image. Then, AlexNet architecture is utilized for training image patches. The Instance-To-Class (ITC) matching technique using K-Nearest Neighbour (KNN) algorithm specifies the class of the facial test image based on patch prediction. The notable contributions of our work are two-folds: the first one is employing ITC technique for patch prediction and the last one is adopting a deep learning technique for feature extraction and handling partial occlusion problem. The achieved accuracies on two de-facto datasets show that our method outperforms several existing methods that use hand-designed feature descriptors.
  PubDate: 2022-12-01
High Sensibility Optical Methane Sensor Based on Insertion of
Cryptophane-E Cavity in 1D Photonic Crystal
- Free pre-print version: Loading...
  Rate this result: What is this?
  
  Please help us test our new pre-print finding feature by giving the pre-print link a rating.
  A 5 star rating indicates the linked pre-print has the exact same content as the published article.
  
  Abstract: In this paper, we proposed a miniaturized, simple, highly sensitive, and high-precision gas sensor for the measurement of concentration of methane. This gas sensor is based on the inclusion of a cryptophane-E cavity in a one-dimensional perfect photonic crystal (PC) composed by alternating layers of Silicon (Si) and Air. The detection principle of this sensor based on the variation of the refractive index (RI) of cryptophane E due to a change in the concentration of methane which induces a shift in resonant wavelength of the cavity (cavity states) in the band gaps, allowing precision and efficient measurement of methane concentration. The band structure and the transmission spectrum are both calculated by the Green function method (GFM). Numerous geometrical and physical parameters like the thickness of the cavity layer and the concentration of methane gas are properly optimized to envisage high sensing performances. The numerical results show that the photonic cavity state (defect mode), which appears in the band gap, is caused by the infiltration of methane into the cryptophane E middle layer. This cavity state can be used for detection purposes in environmental monitoring. The cavity state wavelength is sensitive to the cryptophane E-methane mixture and a variation in the refractive index as ∆n = 10–3 can be detected. The limit of detection value of the proposed sensor is approximately 10–3 refractive index unit, which is very low, as is always expected for chemical sensing designs. This system could be employed for monitoring in the environmental field, for detection of dangerous and/or air polluting gas concentrations, and for liquid analysis with excellent performance.
  PubDate: 2022-12-01
Neural Network: Predator, Victim, and Information Security Tool
- Free pre-print version: Loading...
  Rate this result: What is this?
  
  Please help us test our new pre-print finding feature by giving the pre-print link a rating.
  A 5 star rating indicates the linked pre-print has the exact same content as the published article.
  
  Abstract: The article deals with information security problems associated with neural networks. Malicious neural networks, attacks on neural networks, the use of neural networks as an information security tool, and neural network attack tools are considered. Methods for improving the information security of systems that include neural network components are proposed.
  PubDate: 2022-12-01
Survey on Computational Techniques for Pigmented Skin Lesion Segmentation
- Free pre-print version: Loading...
  Rate this result: What is this?
  
  Please help us test our new pre-print finding feature by giving the pre-print link a rating.
  A 5 star rating indicates the linked pre-print has the exact same content as the published article.
  
  Abstract: Skin lesion segmentation is the first step in skin lesion assessment, and it can help with the following classification task. It is a complex job because the borders of pigment regions may be hazy and the whole lesion can sometimes be the same color. Therefore, this paper provides a comprehensive review of the methodologies proposed in research papers published since 2017. Concerning the image segmentation task, the techniques were classified based on their principle of operation. The works are looked at with the help of certain datasets, year-by-year publications, performance evaluation measures, and other tools. The survey concluded that many researchers have successfully highlighted image segmentation of skin lesions; moreover, improved methods are needed to enhance the performance.
  PubDate: 2022-12-01
Introductory Review on All-Optical Machine Learning Leap in Photonic
Integrated Circuits
- Free pre-print version: Loading...
  Rate this result: What is this?
  
  Please help us test our new pre-print finding feature by giving the pre-print link a rating.
  A 5 star rating indicates the linked pre-print has the exact same content as the published article.
  
  Abstract: The human brain is the most complex circuit on the planet and the circuits inspired by the operation of the biological neuron are the most desired computing need. Artificial neural networks (ANN) are circuits that can replicate the biological neuron. Optical computing already doing wonders in integrated circuit technology and therefore the photonic implementation of neural networks is one of the most appealing technologies of the current era due to its low power consumption and high bandwidth. The ANN models are designed as per the signal processing of the human brain therefore they can be used to improve the analytic power of any system. This article reviews the advancement in optical neural networks and their application for future perspective.
  PubDate: 2022-12-01
Use of Neural Networks and Decision Trees in Compression of 2D and 3D
Digital Signals
- Free pre-print version: Loading...
  Rate this result: What is this?
  
  Please help us test our new pre-print finding feature by giving the pre-print link a rating.
  A 5 star rating indicates the linked pre-print has the exact same content as the published article.
  
  Abstract: The article considers a compression framework for 2D/3D digital signals, including digital images and hyperspectral data. A compression framework is proposed that uses neural networks to exclude insignificant signal portions from the compression process with restoring these portions in decompression. Neural networks for key compression operations (sample prediction, determination of insignificant portions in decompression, etc.) are chosen. Additionally, the machine learning algorithms are modified to incorporate into the compression framework. The framework is extended to the case of 3D signals of hyperspectral data. The efficiency of the approach is tested by computer experiments using real 2D and 3D digital data. The experiment proves high efficiency of the compression algorithm and potentiality of using it in data storing and processing systems.
  PubDate: 2022-12-01
Features of the Optical Vortices Diffraction on Silicon Ring Gratings
- Free pre-print version: Loading...
  Rate this result: What is this?
  
  Please help us test our new pre-print finding feature by giving the pre-print link a rating.
  A 5 star rating indicates the linked pre-print has the exact same content as the published article.
  
  Abstract: We study the diffraction of optical vortices on silicon ring gratings and diffractive axicons with a change in the relief height of optical elements in this paper. It was shown that for silicon ring gratings with variable height it is possible to obtain a light focal needle 83.8% longer than the light needle formed by a standard diffractive axicon. The increasing of the diffractive axicon height reduces the size of the focal spot by 24% compared to an axicon with a standard height.
  PubDate: 2022-12-01
  DOI: 10.3103/S1060992X22050095
Coplanarity-Based Approach for Camera Motion Estimation Invariant to the
Scene Depth
- Free pre-print version: Loading...
  Rate this result: What is this?
  
  Please help us test our new pre-print finding feature by giving the pre-print link a rating.
  A 5 star rating indicates the linked pre-print has the exact same content as the published article.
  
  Abstract: In this paper, we propose a method for estimating the parameters of camera movement from images obtained from this camera. This method is equally effectively applicable to flat and three-dimensional scenes. The proposed method allows avoiding the restrictions imposed on the set of initial data when using the fundamental matrix and the projective transformation matrix.
  PubDate: 2022-12-01
  DOI: 10.3103/S1060992X22050058
Automatic Dominant Orientation Estimation in Texture Images Using the
Scattering Ellipse of the Gradients
- Free pre-print version: Loading...
  Rate this result: What is this?
  
  Please help us test our new pre-print finding feature by giving the pre-print link a rating.
  A 5 star rating indicates the linked pre-print has the exact same content as the published article.
  
  Abstract: In this paper, the problem of estimating the dominant image orientation (DIO) is experimentally investigated using the characteristics of the scattering ellipse of the gradient field components. According to these characteristics, the angle of rotation, the parameters of the shape of the Weibull distribution characterizing the blurring of the image, as well as the assessment of the sharpness of the DIO angle determination, are estimated. The results of an experimental study of the properties of DIO in various situations, such as the implementation of rotations of texture and other types of images by various methods, the presence of distorting noise, etc. are given. Using the Photoshop software system, the results of applying the previously proposed approach to assessing the quality of image rotation algorithms called the “backlash method”, are studied.
  PubDate: 2022-12-01
  DOI: 10.3103/S1060992X22050022
Semantic Segmentation of Hyperspectral Imaging Using Convolutional Neural
Networks
- Free pre-print version: Loading...
  Rate this result: What is this?
  
  Please help us test our new pre-print finding feature by giving the pre-print link a rating.
  A 5 star rating indicates the linked pre-print has the exact same content as the published article.
  
  Abstract: using neural networks in hyperspectral imaging helps to get through the obstruction to solving data analysis, classification, and segmentation problems. There are problems, such as vegetations analysis in agriculture, which cannot be solved using classic RGB images due to lack of information. Applying neural networks to hyperspectral images is a sophisticated problem. The aim of this study is to examine concerns about using convolutional neural networks for the semantic segmentation of hyperspectral data. The following problems were considered: large spatial resolution, the influence of neural network’s input size on accuracy and performance; hyperspectral data preprocessing, the influence of dimensionality reduction and brightness equalization; neural network architecture influence on analyzing hyperspectral imaging. Also, the accuracy of neural networks was compared to classic approaches: multinominal logistic regression, random forest algorithm, discriminant analysis. As the result of the study the importance of choosing neural network’s architecture and hyperspectral data preprocessing methods are discussed.
  PubDate: 2022-12-01
  DOI: 10.3103/S1060992X22050071
Integrated Resonant Diffraction Gratings for Bloch Surface Waves
- Free pre-print version: Loading...
  Rate this result: What is this?
  
  Please help us test our new pre-print finding feature by giving the pre-print link a rating.
  A 5 star rating indicates the linked pre-print has the exact same content as the published article.
  
  Abstract: We propose and theoretically and numerically investigate resonant subwavelength “on-chip” diffraction gratings operating with Bloch surface waves (BSWs), which propagate along an interface separating an all-dielectric photonic crystal and a dielectric medium. The investigated structures correspond to periodically arranged dielectric “posts” on the propagation interface. We show that in the regime, in which there is no unwanted scattering of the incident BSW away from the propagation interface, and no excitation of the cross-polarized reflected and transmitted BSWs, in the proposed structures, there exist high-Q resonances and bound states in the continuum, which makes them promising for the utilization as narrowband spectral or spatial (angular) filters for BSWs and as analog optical differentiators and integrators.
  PubDate: 2022-12-01
  DOI: 10.3103/S1060992X22050034
Smartphone Camera Self-Calibration Based on Sensors Reading Consistency
- Free pre-print version: Loading...
  Rate this result: What is this?
  
  Please help us test our new pre-print finding feature by giving the pre-print link a rating.
  A 5 star rating indicates the linked pre-print has the exact same content as the published article.
  
  Abstract: In large-scale series production the time for evaluating the camera spectral sensitivity is strongly limited and measured in units of seconds because of production and economic constraints. To estimate variation of spectral sensitivity properties, manufacturers usually precisely measure only a few sensors (the golden set) and use these measurements to perform quick estimation of any other sensor in the released pack. The main drawback of this approach is that the worst color reproduction error cannot be controlled for a particular device: instability of device production process usually causes significantly different sensors, which may not be included in the golden set. In that case the camera will work with low accuracy during the lifetime. To overcome this problem, we consider a new approach to camera spectral sensitivity estimation during its operation. The main idea is based on consistency estimation of images and average scenes spectra. Users receive such a combination of data in practice, for instance modern phone devices have built-in integral spectrometers. Also, the proposed approach can be considered in the scope of classical problem statement of spectral sensitivity estimation with color charts. In the paper we investigated the accuracy of the method of spectral sensitivity estimation based on the basis calculation with singular value decomposition of the sensitivities from the golden set in combination with different types of regularization.
  PubDate: 2022-12-01
  DOI: 10.3103/S1060992X22050083
Recognition of Half-Integer Order Vortex Beams Using Convolutional Neural
Networks
- Free pre-print version: Loading...
  Rate this result: What is this?
  
  Please help us test our new pre-print finding feature by giving the pre-print link a rating.
  A 5 star rating indicates the linked pre-print has the exact same content as the published article.
  
  Abstract: In this paper, we investigate the possibility of determining the order (both whole and half-integer) of an optical vortex beams based on the application of convolutional neural networks. To improve recognition results, we use paired patterns of vortex beams: intensity distribution in the focal plane with and without astigmatic transformation.
  PubDate: 2022-12-01
  DOI: 10.3103/S1060992X22050046
Computer Simulation of Image Formation by Diffraction Lens
- Free pre-print version: Loading...
  Rate this result: What is this?
  
  Please help us test our new pre-print finding feature by giving the pre-print link a rating.
  A 5 star rating indicates the linked pre-print has the exact same content as the published article.
  
  Abstract: We offer computer technology for simulating the imaging process using diffraction optical harmonic lenses. The technology is built in a paraxial approximation using the laws of geometric optics. To simulate the refractive index versus wavelength, we use a multispectral image with a uniform wavelength distribution in the optical range. To assess the quality of the generated images, the components of the input and output multispectral images are integrated using standard colorimetric observer functions, specified on a discrete set of wavelengths. An illustrative example of computer simulating is given for the case when a color image at the input is presented as a plane perpendicular to the lens optical axis.
  PubDate: 2022-12-01
  DOI: 10.3103/S1060992X2205006X
Simple LASER Tracking Algorithm Using Programmable System on Chip (PSoC)
for Visible Light Communication (VLC)
- Free pre-print version: Loading...
  Rate this result: What is this?
  
  Please help us test our new pre-print finding feature by giving the pre-print link a rating.
  A 5 star rating indicates the linked pre-print has the exact same content as the published article.
  
  Abstract: This paper presents a simple LASER tracking algorithm that can be used for Visible Light Communication (VLC) which is a new technology that uses the visible light (380–780 nm) as a carrier for the data. Recently, VLC has attracted a high attention because it has the potential transmissions at very-high data rates. To attain a good communication in VLC, both the transmitter and receiver must be on one line, otherwise the whole system will fail. The proposed algorithm has been implemented using Two-Dimensional Position Sensitive Detector (PSD) and Programmable System on Chip 3 (PSoC3). The PSD is an optoelectronic sensor which is used to measure the radiation spot position with an ability to sense the position in one or two dimensions. The used PSD sensor consists of two-dimensional array of photodiodes and it has four output photocurrents which vary with the position of incident radiation on the surface. To validate the proposed algorithm, MATLAB is used to simulate the PSD using its modeling equations. Solidworks software tool is also used to design and simulate the gimbal. The hardware implementation is done for the validation purpose by using the PSoC which occupies minimum board space, consumes less power, provides high efficiency and reduces system cost.
  PubDate: 2022-09-01
  DOI: 10.3103/S1060992X22030079
Tunable Narrow Filter Based on 1D Photonic Star Waveguides Containing
Metamaterial Resonators Defects for Frequency Division Multiplexing
Applications
- Free pre-print version: Loading...
  Rate this result: What is this?
  
  Please help us test our new pre-print finding feature by giving the pre-print link a rating.
  A 5 star rating indicates the linked pre-print has the exact same content as the published article.
  
  Abstract: In the framework of the Green Function method (GFM), we investigate the transmission spectrum and the band structure of the one-dimensional perfect photonic star waveguides (SWGs) with the left-handed materials (LHM). This perfect structure is composed of backbones' periodicity that is characterized by positive magnetic permeability μ and dielectric permittivity ε, grafted in its extremity by resonators of negative permittivity and magnetic permeability (two kinds of single-negative) for a frequency range specify well. This photonic structure exhibits passbands separated by large photonic bandgaps (PBGs), which are due to the periodicity of the structure and the resonance modes of the resonators. Inside this perfect structure, we consider geometrical defects at the resonators level located in two different positions. The presence of such defects in this structure creates inside the photonic band gaps of the perfect structure, very narrow filtered frequencies (defects modes) with a significant quality factor Q and a very high transmission rate. Some defects modes shift towards the low frequencies when the defects lengths increase, while other defects modes shift towards high frequencies, which is completely different when considering this kind of defectives structure with the presence of right-handed materials (RHM). The results show very clearly that for very specific values of the lengths of the two defects, two defects’ modes interplay with each other, after this interaction, we obtain the phenomenon of behavior change between two defects modes (energy exchange). We can claim that our LHM filter is suitable for frequency selection, noise filtering, and FDM (Frequency Division Multiplexing) electromagnetic communication applications.
  PubDate: 2022-09-01
  DOI: 10.3103/S1060992X22030031
Weakly Labeled Semi-Supervised Sound Event Detection Based on
Convolutional Independent Recurrent Neural Networks
- Free pre-print version: Loading...
  Rate this result: What is this?
  
  Please help us test our new pre-print finding feature by giving the pre-print link a rating.
  A 5 star rating indicates the linked pre-print has the exact same content as the published article.
  
  Abstract: Sound Event Detection (SED) needs to identify the sound events in a recording and detect the onset and offset times of them. Deep learning-based methods such as Recurrent Neural Networks (RNN) achieved state-of-the-art results in sound event detection. However, two open challenges still remain: performance was severely degraded in real-life environments and frame-level labeled prediction using few clip-level labeled audio data is difficult. To solve the above two problems, we proposed a weakly labeled semi-supervised method (named CIRAS) for SED, which can be easily trained from noisy data in a weakly semi-supervised fashion, requiring only clip-level labeled audio data and unlabeled audio data. Firstly, we constructed a Gated Convolutional Bidirectional Independent Recurrent Neural Network (GCBIndRNN), the RNN residual connections over layers to increase the depth of the network, which effectively alleviates the vanishing and exploding gradient in the backpropagation. Secondly, we introduced an Efficient Channel Attention (ECA), which enables the network to pay more attention to the sound event feature under the background noise. Finally, we used the Sample Relation Consistency Mean Teacher (SRC-MT) semi-supervised framework, which can be efficiently utilized clip-level labeled audio data and unlabeled audio data to acquire frame-level label prediction. Our proposed method is evaluated on task 4 of the DCASE 2018 and 2019 challenge and compared with several baselines, demonstrating competitive results in terms of F1-score.
  PubDate: 2022-09-01
  DOI: 10.3103/S1060992X22030110
Comparative Analysis of a Neural Network with Calculated Weights and a
Neural Network with Random Generation of Weights Based on the Training
Dataset Size
- Free pre-print version: Loading...
  Rate this result: What is this?
  
  Please help us test our new pre-print finding feature by giving the pre-print link a rating.
  A 5 star rating indicates the linked pre-print has the exact same content as the published article.
  
  Abstract: The paper discusses the capabilities of multilayer perceptron neural networks implementing metric recognition methods, for which the values of the weights are calculated analytically by formulas. Comparative experiments in training a neural network with pre-calculated weights and with random initialization of weights on different sizes of the MNIST training dataset are carried out. The results of the experiments show that a multilayer perceptron with pre-calculated weights can be trained much faster and is much more robust to the reduction of the training dataset.
  PubDate: 2022-09-01
  DOI: 10.3103/S1060992X22030080
Neural Networks for Classification and Unsupervised Segmentation of
Visibility Artifacts on Monocular Camera Image
- Free pre-print version: Loading...
  Rate this result: What is this?
  
  Please help us test our new pre-print finding feature by giving the pre-print link a rating.
  A 5 star rating indicates the linked pre-print has the exact same content as the published article.
  
  Abstract: For computer vision systems of autonomous vehicles, an important task is to ensure high reliability of visual information coming from on-board cameras. Frequent problems are contamination of the camera lens, its defocusing due to mechanical damage, image motion blur in low light conditions. In our work, we propose a novel neural network approach to the classification and unsupervised segmentation of visibility artifacts on monocular camera images. It is based on the compact classification deep neural network with an integrated modification of the gradient method for class activation map and segmentation mask generating. We present a new dataset named Visibility Artifacts containing over 22 300 images including six common artifacts: complete loss of camera visibility, strong or partial contamination, rain or snow drops, motion blur, defocus. To check the quality of artifact localization, a small test set with ground truth masks is additionally labeled. It allowed us to objectively quantitatively compare various methods for constructing class activation maps (CAMERAS, FullGrad, original and modified Grad-CAM, Layer-CAM), which demonstrated image segmentation quality above 54% mIoU without any supervision. This is a promising result. Experiments with the developed dataset demonstrated the superiority of the neural network classification method ResNet-18_U (with test accuracy of 99.37%), compared to more complex convolutional (ResNet-34, ResNeXt-50, EfficientNet-B0) and transformer (ViT-Ti, DeiT-Ti) neural networks. The code of the proposed method and the dataset are publicly available at https://github.com/vd-kuznetsov/CaUS_Visibility_Artifacts.
  PubDate: 2022-09-01
  DOI: 10.3103/S1060992X22030043