Computer Optics

864-871

Authors: V.V. Kotlyar, A.A. Kovalev

Here, we investigate coaxial superpositions of Gaussian optical vortices that can be described by a geometric sequence. For all superpositions analyzed, a topological charge (TC) is derived. In the initial plane, the TC can be either integer or half-integer, acquiring an integer value upon free-space propagation of the light field. Generally, the geometric sequence of optical vortices (GSOV) has three integer parameters and one real parameter. Values of these four parameters define the TC of the GSOV. Upon free-space propagation, the intensity pattern of the GSOV is not conserved, but can have intensity petals whose number is equal to one of the four beam parameters. If the GSOV has a unit real parameter, all constituent angular harmonics in the superposition have the same weight. In this case, the TC of the superposition is equal to the average index of the constituent angular harmonics. For instance, if the TC of the first and of the last angular harmonics, respectively, equals k and n, then the total TC of the superposition in the initial plane will be (n + k) /2, becoming equal to n upon free-space propagation.

877-883

Authors: N.G. Stsepuro, M.S. Kovalev, G.K. Krasin, I.V. Gritsenko, A.V. Bobkov, S.I. Kudryashov

The transport-of-intensity equation provides a new non-interferometric and non-iterative access to quantitative information about the phase of a light wave by measuring intensity distributions. This equation can be used to implement a simple and accurate spatial phase measurement for optical testing of spherical surfaces. The method requires only a CMOS camera, which records transverse field intensity distributions in several planes. Processing of experimental measurements with specialized software allows one to reconstruct the value of the radius of curvature of the spherical surface under test with high accuracy. The method is compared with measurements made by an interferometer, showing the difference between the values of the surface radius of curvature to be 0.01 % or less and indicating good agreement of the results.

892-898

Authors: G.I. Greisukh, E.G. Ezhov, I.A. Levin

The possibility of achieving superior optical characteristics and passive athermalization in dual-band IR objectives of a simple design is shown. This is ensured, in particular, by using refractive lenses made of chalcogenide glasses in the optical scheme. Recommendations regarding the layout of the optical scheme and obtaining initial design parameters for the subsequent optimization are given. The reliability and effectiveness of the recommendations are confirmed by the results of designing a super-aperture refractive-lens objective operating in the mid- and long-IR subranges and forming a high-quality polychromatic image on the matrix of an uncooled microbolometer in the operating temperature range from – 40°C to + 60°C.

905-913

Authors: V.M. Kotov, S.V. Averin, A.A. Zenkina, A.S. Belousova

Properties of a two-channel acousto-optic (AO) spatial frequency filter deflecting optical beams in two symmetrical orders are studied. It is found that the non-uniform distribution of the field in each order makes it possible to process images in two channels using different transfer functions. The formation of a two-dimensional contour in two diffraction orders is confirmed ex-perimentally on the basis of image processing for a wavelength of 0.63 μm. A TeO 2 AO cell oper-ating at an acoustic frequency of 26 MHz is used as a two-channel spatial frequency filter.

921-928

Authors: O.S. Seredin, O.A. Kushnir, S.A. Fedotova

The study is a comparative analysis of two fast reflection symmetry axis detection methods: an algorithm to refine the symmetry axis found with a chain of skeletal primitives and a boundary method based on the Fourier descriptor. We tested the algorithms with binary raster images of plant leaves (FLAVIA database). The symmetry axis detection quality and performance indicate that both methods can be used to solve applied problems. Neither method demonstrated any significant advantage in terms of accuracy or performance. It is advisable to integrate both methods for solving real-life problems.

939-947

Authors: S.V. Eremeev, A.V. Abakumov, D.E. Andrianov, D.V. Titov

A new method for decomposing an image into separate objects of interest is proposed in the article. The developed method is based on the use of persistent homology. A process of direct and reverse image transformation is shown. Following direct transformation, the original image is represented as a set of matrices that can be divided into basic and detailing ones. The basic matrices contain information about the basic structure of objects in the images, and the detailing ones include data about the details of these objects, about small objects or the noise component. It is shown that there is a certain analogy with the Wavelet transformation, but the proposed method is based on a fundamentally different theoretical basis. A numerical example reflecting the basic essence of the method is described in detail. Properties of the decomposition and the possibility of using standard algebraic operations on decomposition matrices are described. The reverse transformation allows us to take into account the changed properties of individual objects and synthesize a new image. Prospects of using the proposed decomposition for solving practical problems are demonstrated. Algorithms have been developed for binarization of images and removal of text on a non-uniform background. Data analysis and processing is carried out using a unified approach in the space of decomposition matrices. The results of binarization have shown that, when compared with analogues, the developed algorithm will perform better when the binarization is used to isolate a multitude of individual objects. The obtained results of the algorithm for deleting text on a non-uniform background confirm that the information is completely deleted without affecting the rest image areas.

955-962

Authors: A.A. Axyonov, D.A. Ryumin, A.M. Kashevnik, D.V. Ivanko, A.A. Karpov

The paper proposes a method of visual analysis for automatic speech recognition of the vehicle driver. Speech recognition in acoustically noisy conditions is one of big challenges of artificial intelligence. The problem of effective automatic lip-reading in vehicle environment has not yet been resolved due to the presence of various kinds of interference (frequent turns of driver's head, vibration, varying lighting conditions, etc.). In addition, the problem is aggravated by the lack of available databases on this topic. A MediaPipe Face Mesh is used to find and extract the region-of-interest (ROI). We have developed End-to-End neural network architecture for the analysis of visual speech. Visual features are extracted from a single image using a convolutional neural network (CNN) in conjunction with a fully connected layer. The extracted features are input to a Long Short-Term Memory (LSTM) neural network. Due to a small amount of training data we proposed that a Transfer Learning method should be applied. Experiments on visual analysis and speech recognition present great opportunities for solving the problem of automatic lip-reading. The experiments were performed on an in-house multi-speaker audio-visual dataset RUSAVIC. The maximum recognition accuracy of 62 commands is 64.09 %. The results can be used in various automatic speech recognition systems, especially in acoustically noisy conditions (high speed, open windows or a sunroof in a vehicle, backgoround music, poor noise insulation, etc.) on the road.

971-979

Authors: D. Kaibassova, М. Nurtay, А. Tau, М. Kissina

This study is devoted to the application of fine-tuning methods for Transfer Learning models to solve the multiclass image classification problem using the medical X-ray images. To achieve this goal, the structural features of such pre-trained models as VGG-19, ResNet-50, InceptionV3 were studied. For these models, the following fine-tuning methods were used: unfreezing the last convolutional layer and updating its weights, selecting the learning rate and optimizer. As a dataset chest X-Ray images of the Society for Imaging Informatics in Medicine (SIIM), as the leading healthcare organization in its field, in partnership with the Foundation for the Promotion of Health and Biomedical Research of Valencia Region (FISABIO), the Valencian Region Medical ImageBank (BIMCV) ) and the Radiological Society of North America (RSNA) were used. Thus, the results of the experiments carried out illustrated that the pre-trained models with their subsequent tuning are excellent for solving the problem of multiclass classification in the field of medical image processing. It should be noted that ResNet-50 based model showed the best result with 82.74 % accuracy. Results obtained for all models are reflected in the corresponding tables.

988-1019

Authors: I.A. Hodashinsky, K.S. Sarin, M.B. Bardamova, M.O. Svetlakov, A.O. Slezkin, N.P. Koryshev

A review of noninvasive biometric methods for detecting and predicting neurodegenerative diseases is presented. An analysis of various modalities used to diagnose and monitor diseases is given. Such modalities as handwritten data, electroencephalography, speech, gait, eye movement, as well as the use of compositions of these modalities are considered. A detailed analysis of modern methods and solutions based on machine learning is conducted. Data sets, preprocessing methods, machine learning models, and accuracy estimates for disease diagnosis are presented. In the conclusion current open problems and future prospects of research in this direction are considered.

682-691

Authors: S.V. Karpeev, V.V. Podlipnov, S.A. Degtyarev, A.M. Algubili

Complex polarization-phase transformations are considered, which are implemented using easy-to-manufacture optical elements. The manufacturing technology of such elements is based on the axially symmetric discretization of the required polarization and phase distributions. This representation leads to optical elements in the form of sector sandwich structures consisting of polarizing and phase plates stacked together. The paper numerically and experimentally investigates the main types of such sector sandwich structures for the formation of second-order cylindrical polarizations.

701-712

Authors: C.N. Alexeyev, S.S. Aliyeva, E.V. Barshak, B.P. Lapin, M.A. Yavorsrky

In this paper, we studied transformations of structured light and its angular momentum in a three-coil optical snake – a coil resonator composed of 3 evanescently uniformly coupled coils of a multimode fiber. We have suggested a fully vectorial theory of normal modes of the 3-coil resona-tor, which takes account of the spin-orbit interaction. On the basis of the analytical expressions for such normal modes and their propagation constants we have studied transmission of some types of structured light beams – optical vortices, Hermite-Gaussian-like and Laguerre-Gaussian beams – through such a system. We have shown the possibility of a super-efficient parametric control over the topological charge, orbital and spin angular momenta of the outcoming optical field by this system. We have theoretically demonstrated implementation of logic X and Y Pauli gates for light beams carrying orbital angular momentum on the basis of such a 3-coil resonator.

724-732

Authors: A.O. Frolov, А.V. Ustinov, S.N. Khonina

In this paper, we study a change in the propagation trajectory of a set of autofocusing laser beams using a fractional Fourier transform. Clusters of displaced bounded Airy-Gaussian beams supplemented by a phase function that deflects the beam similarly to a prism are considered. The shift and phase deviation (according to the carrier spatial frequencies) make it possible to change the propagation trajectory of a set of autofocusing beams. The influence of the parameters under consideration on the properties of autofocusing of a cluster of Airy-Gaussian beams is investigated by means of numerical simulation.

741-751

Authors: S.I. Kharitonov, N.L. Kazanskiy, S.G. Volotovsky, S.N. Khonina

The work is devoted to the development of the quantum theory of diffractive optical elements. Aspects of quantum optics are considered by the example of light diffraction from a dielectric plate in a resonator. The paper shows the connection between the classical and quantum solution of the problem of diffraction by a dielectric plate. Expressions are obtained for the eigenmodes of such a resonator, as well as for the operators of the vector magnetic potential and the electric field strength. The method proposed in this paper can be easily extended to dielectric plates with a diffractive microrelief, that is, to diffractive optical elements.

763-773

Authors: N.N. Vasilyuk

A method for calculating of a blur kernel arising from the rotation of a digital camera is proposed. The rotation is measured with a three-axis gyroscope attached to the camera. Differential equations of a continuous trajectory of the rotational blur starting from a selected pixel are obtained. These equations are presented both in the form of an explicit system of differential equations for the blur curve in the focal plane and in the form of a matrix equation for the increment of the camera attitude. An expression is given for the integral of the energy illumination from a point light source along the continuous blur trajectory. The integral takes into account the point spread function and the aperture functions of individual photosensitive cells of the photodetector array. The calculation of the integral values for all photosensitive cells illuminated with the point source gives a discrete kernel of rotational blur starting at the selected pixel. Algorithms for the numerical integration of the blur equations are described. The analysis of the blur equations are carried out, characteristic features of the kernels are highlighted and their non-homogeneity is shown, with the kernels of rotational blur revealed not to coincide with each other for different pixels. An example of the synthesis of the blur kernels for given rotation parameters of the digital camera is given.

783-789

Authors: S.H. Jeevith, S. Lakshmikanth

In various computer vision applications, the moving object detection is an essential step. Principal Component Analysis (PCA) techniques are often used for this purpose. However, the performance of this method is degraded by camera shake, hidden moving objects, dynamic background scenes, and / or fluctuating exposure. Robust Principal Component Analysis (RPCA) is a useful approach for reducing stationary background noise as it can recover low rank matrices. That is, moving object is formed by the low power models and the static background of RPCA. This paper proposes a simple alternative minimization algorithm to fix minor discrepancies in the original Principal Component Pursuit (PCP) or RPCA function. A novel hybrid method of cartoon texture features used as a data matrix for RPCA taking into account low-ranking and rare matrix is presented. A new non-convex function is proposed to better control the low-range properties of the video background. Simulation results demonstrate that the proposed algorithm is capable of giving consistent random estimates and can indeed improve the accuracy of object recognition in comparison with existing methods.

801-807

Authors: A.D. Sokolova, A.V. Savchenko, S.I. Nikolenko

One of main issues in face identification is the lack of training data of specific type (bad quality image, varying scale or illumination, children/old people faces, etc.). As a result, the recogni-tion accuracy may be low for input images which are not similar to the majority of images in the dataset used to train the feature extractor. In this paper, we propose that this issue is dealt with by the automatic detection of such out-of-distribution data based on the addition of a preliminary stage of their automatic rejection using a special convolutional network trained using a set of rare data collected using various transformations. To increase the computational efficiency, the decision about the presence of a rare image is made on the basis of the same face descriptor that is used in the classifier. Experimental research confirmed the accuracy improvement of the proposed approach for several datasets of faces and modern neural network descriptors.

818-827

Authors: D.E. Plotnikov, P.A. Kolbudaev, E.A. Loupian

The paper describes a method for automatic subpixel-accurate geographic referencing of imagery acquired by KMSS-M with 60 m spatial resolution, based on leveraging a coarse, reconstructed, cloud-free and daily updated MODIS surface reflectance reference image. The method is based on maximizing Pearson's correlation value when determining an optimal local displacement of the distorted image fragment by comparing with the reference image. To assess the effectiveness of the method when used over continental-scale and heterogeneous areas, three experiments were carried out providing quantitative estimates of imagery registration errors: an experiment with model datasets, an experiment to estimate the absolute registration error of MODIS reference imagery, and an experiment to estimate the registration error of geocorrected KMSS-M data. Experimental evaluation of the method based on model datasets of decameter-resolution Sentinel-2 (MSI) imagery demonstrated its robustness when used over a variety of environmental conditions over a one year-long observation period. The average georeferencing error of MODIS coarse-resolution reference was shown to be less than 20 meters in Red and Near-infrared bands. Corrected KMSS-M imagery evaluation over the Russian Grain Belt within 2020 has shown, on average, the subpixel referencing accuracy both in Red and Near-infrared bands, while the average absolute georeferencing error of the original uncorrected KMSS-M imagery was shown to be about 3 kilometers. Subpixel registration accuracy of KMSS-M imagery, corrected with MODIS-based coarse-resolution reference, opens new prospects for using multi-temporal analysis of this multispectral surface reflectance data in a variety of scientific and practical applications associated with vegetation cover satellite monitoring. The technological flexibility of the method ensures its applicability to data from other satellite systems for Earth optical remote sensing.

840-847

Authors: Е.E. Aydarkin, N.S. Mogilevskaya

The paper investigates the ability of the equal-weight columns method to resist grouping erasures. The Gilbert model for error generation flow is adapted for the case of erasures. A simulation model of the erasure-correction channel is constructed with the possibility of choosing the type of erasure and the method of protection against erasure. With the help of this model, an experimental study of the equal-weight column method and its modifications is conducted and a detailed analysis of the results with conclusions for developers of networks and data transmission channels is carried out. An estimate of the decoding probability is constructed. A method of dealing with clustering erasures by using additional redundancy is proposed.

522-530

Authors: V.V. Kotlyar, E.G. Abramochkin, A.A. Kovalev, A.G. Nalimov

It is shown theoretically that an astigmatic transformation of an edge dislocation (straight line of zero intensity) of the ν-th order (ν=n+α is a real positive number, n is integer, 0<α<1 is the fractional part of the number) forms at twice the focal length from a cylindrical lens n optical elliptical vortices (screw dislocations) with a topological charge of –1, located on a straight line perpendicular to the edge dislocation. Coordinates of these points are zeros of the Tricomi function. At some distance from these vortices and on the same straight line, another additional vortex with a topological charge of –1 is also generated, which moves to the periphery if α decreases to zero, or approaches n vortices if α tends to 1. In addition, at the periphery in the beam cross-section, a countable number of optical vortices (intensity zeros) are formed, all with a topological charge of –1, which are located on diverging curved lines (such as hyperbolas) equidistant from a straight line on which the main n intensity zeros are located. These "accompanying" vortices approach the center of the beam, following the additional "passenger" vortex, if 0<α<0.5, or move to the periphery, leaving the "passenger" next to the main vortices, if 0.5<α<1. At α=0 and α=1, the "accompanying" vortices are situated at infinity. The topological charge of the entire beam at fractional ν is infinite. The numerical simulation confirms theoretical predictions.

537-546

Authors: D.A. Savelyev

The focusing features of a super-Gaussian beam, as well as Laguerre-super-Gaussian (1,0) modes with radial and circular polarizations on ring gratings (direct and inverse) with a variable height of individual relief rings were investigated in this paper. The change in the height of the relief from the maximum height in the center to the minimum at the edges of the element (direct ring grating) is considered and the reverse case, when the maximum height of the relief was at the edges and the minimum in the center (reverse ring grating). The comparison was carried out with the action of a diffractive axicon with a comparable grating period. Numerical simulation was carried out by the finite differences in the time domain method. It was shown that the direction of change in the height of the relief rings of elements significantly affects the diffraction pattern in the near zone. In particular, for a super-Gaussian beam, the use of a direct ring grating made it possible to reduce the size of the focal spot by 18.7% for radial polarization ('sigma'=5 μm), the use of a reverse ring grating led to a decrease in the focal spot by 36.9% ('sigma'=12.25 µm) in comparison with the action of a diffractive axicon. It was also shown in the paper that when the reverse ring grating is illuminated by the Laguerre-superGauss (1,0) mode with circular polarization, the formation of a narrow and extended region of reduced intensity on the optical axis (optical trap) is observed.

555-560

Authors: V.S. Petrakova, A.S. Tsipotan, V.V. Slabko

One approach for the formation of structures with complex geometries at the nanoscale is the step-by-step assembly. In this case, it is necessary to be able to estimate the time required to establish orientational equilibrium for a preformed pair of particles. This process is statistical in nature and depends on the mechanism of interaction of the ensemble with the external field. The orientation of particles in an alternating field is associated with certain relaxation times, which depend on the viscosity and temperature of the medium, as well as on the geometric structure of the samples. This paper proposes an mathematical model of the process of establishing the distribution of nanoparticles pairs orientations taking into account the friction force, thermal motion, and the orienting laser field. A statistical orientation distribution was obtained for CdTe particles in the field of moderate laser radiation, and the average time for establishing orientational equilibrium was estimated.

567-589

Authors: V.V. Arlazarov, E.I. Andreeva, K.B. Bulatov, D.P. Nikolaev, O.O. Petrova, B.I. Savelev, O.A. Slavin

This paper analyzes the problems of document image recognition and the existing solutions. Document recognition algorithms have been studied for quite a long time, but despite this, currently, the topic is relevant and research continues, as evidenced by a large number of associated publications and reviews. However, most of these works and reviews are devoted to individual recognition tasks. In this review, the entire set of methods, approaches, and algorithms necessary for document recognition is considered. A preliminary systematization allowed us to distinguish groups of methods for extracting information from documents of different types: single-page and multi-page, with text and handwritten contents, with a fixed template and flexible structure, and digitalized via different ways: scanning, photographing, video recording. Here, we consider methods of document recognition and analysis applied to a wide range of tasks: identification and verification of identity, due diligence, machine learning algorithms, questionnaires, and audits. The groups of methods necessary for the recognition of a single page image are examined: the classical computer vision algorithms, i.e., keypoints, local feature descriptors, Fast Hough Transforms, image binarization, and modern neural network models for document boundary detection, document classification, document structure analysis, i.e., text blocks and tables localization, extraction and recognition of the details, post-processing of recognition results. The review provides a description of publicly available experimental data packages for training and testing recognition algorithms. Methods for optimizing the performance of document image analysis and recognition methods are described.

596-602

Authors: R.A. Paringer, A.V. Mukhin, N.Y. Ilyasova, N.S. Demin

Advances in the neural networks have brought revolution in many areas, especially those related to image processing and analysis. The most complex is a task of analyzing biomedical data due to a limited number of samples, imbalanced classes, and low-quality labelling. In this paper, we look into the possibility of using neural networks when solving a task of semantic segmentation of fundus. The applicability of the neural networks is evaluated through a comparison of image segmentation results with those obtained using textural features. The neural networks are found to be more accurate than the textural features both in terms of precision (~25%) and recall (~50%). Neural networks can be applied in biomedical image segmentation in combination with data balancing algorithms and data augmentation techniques.

612-620

Authors: A.Y. Denisova, V.V. Fedoseev

The article proposes a feature system designed to detect presentation attacks on facial authentication systems. In this type of attack, an attacker disguises as an authorized user using his image. The feature system assumes the possibility of using one or more special imaging sensors in addition to the basic RGB camera (thermal cameras, depth cameras, infrared cameras). The method has demonstrated a low error rate on the WMCA dataset, while experiments have shown its ability to remain effective in the case of the lack of training data. The comparative experiments carried out showed that the proposed method surpassed the RDWT-Haralick-SVM algorithm, and also approached the results of the MC-CNN algorithm, based on deep learning, which requires a significantly larger amount of training data.

628-633

Authors: V.Y. Kovalev, A.G. Shishkin

In this paper, a multiclass image semantic segmentation problem was solved. For analysis, images of the intracytoplasmic sperm injection process were used. For training the neural network, 656 frames were manually labelled. As a result, each pixel of the images was assigned to one of four classes: microinjector, suction micropipette, oolemma, background. An analysis of modern approaches was carried out and the best architecture, encoders, and hyperparameters of the neural network were selected experimentally: the convolutional neural network FPN (feature pyramid network) with the resnext101 encoder having a depth of 101 layers with 32 parallel separable convolutions. The developed neural network model has allowed obtaining the segmentation efficiency of IOU=0.96 at the algorithm speed of 15 frames per second.

643-649

Authors: M.V. Gashnikov, A.V. Kuznetsov

We investigate algorithms for detecting artificial fragments of remote sensing images generated by adversarial neural networks. We consider a detector of artificial images based on the detection of a spectral artifact of generative-adversarial neural networks that is caused by a layer for enhancing the resolution. We use the detecting algorithm to detect artificial fragments embedded in natural remote sensing images using an adversarial neural network that includes a contour generator. We use remote sensing images of various types and resolutions, whereas the substituted areas, some being not simply connected, have different sizes and shapes. We experimentally prove that the investigated spectral neural network detector has high efficiency in detecting artificial fragments of remote sensing images.

659-665

Authors: A.V. Kamenskiy

The development of modern technologies in the field of image formation leads to an increase in the size of the generated images, as a result the question of reducing the processing computational costs arises, and this is an important factor in the creation of real-time systems. The study provides a description of high-speed recursive-separable filters for improving the quality of images, which, due to the peculiarities of their implementation, can reduce the number of computational operations required for the image processing process. This type of filters is obtained from two-dimensional linear digital filters, which are modified by applying recursive and separable properties to them. The MATLAB environment computing method for implementation of these filters is described. An extensive performance research of the developed filters has been carried out at various sizes of the test image and on various experimental installations. The comparison with the classical two-dimensional convolution method of the developed filters is demonstrated, and it shows the time gain required for the image processing. The results obtained can be applied in biomedical image processing systems or in vision systems working in heavy weather conditions.

366-374

Authors: V.V. Kotlyar, A.A. Kovalev, A.G. Nalimov

Topological charges (TC) of a superposition of two Laguerre-Gauss (LG) beams (0,m) and (0,n) symmetrically shifted from the optical axis are found in this work. It is shown that if m=n, then the TC of the superposition is equal to n. That is, two identical off-axis LG beams have the net TC as that of a single LG beam. If m

381-387

Authors: S.S. Stafeev, V.D. Zaitsev, V.V. Kotlyar

We consider sharp focusing of a linearly polarized light beam. Using the Richards-Wolf formalism, we show that before and after the focal plane there are cross-section regions in which the polarization is circular (elliptical). When passing through the focal plane, the direction of rotation of the polarization vector is reversed. If before the focus the light is left-hand circular polarized at a definite region of the beam cross-section, then exactly at the focus the polarization becomes linear at this cross-section region, before becoming right circular polarized after the focus at this region. This effect allows linearly polarized light to be used to rotate weakly absorbing dielectric microparticles around their center of mass.

395-405

Authors: I.M. Sizova, D.B. Stavrovskii

A theoretical investigation has been carried out of the evolution of a monochromatic laser beam profile after the beam exit from an apodizer. The optical scheme of the apodizer consists of a circular serrated aperture with a narrow opaque corrective ring and a spatial filter based on the Kepler telescope with a pinhole. Characteristic parameters of the apodizer have been found, which allows one to use the corrective ring for the improvement of the apodized beam profile quality and reduce the beam diffraction distortion at beam propagation over distances corresponding to the Fresnel numbers NF~5...2.

415-421

Authors: V.V. Podlipnov, D.A. Bykov, D.V. Nesterenko

CdTe thin films have been deposited by thermal evaporation on heated glass substrates. Structural properties of the CdTe thin films were studied by scanning electron microscopy and Raman spectroscopy. Optical properties were examined by ellipsometry and Fourier spectroscopy. We revealed the low absorption of the synthesized thin films in the infrared (IR) region. We investigated the sensing capability of metal / dielectric / dielectric structures based on a CdTe waveguide layer and gold thick film. The reflectivity spectra of the structures with the water, ethanol, and isopropanol as sensing media exhibit resonance line shapes. The positions of the resonances correspond to those of local maximal values of the solvents absorption. The obtained results can be used for developing the applications of optical resonances in the IR region.

429-435

Authors: A. Sheshkus, A. Chirvonaya, V.L. Arlazarov

In this paper, we study the problem of feature points description in the context of document analysis and template matching. Our study shows that specific training data is required for the task especially if we are to train a lightweight neural network that will be usable on devices with limited computational resources. In this paper, we construct and provide a dataset of photo and synthetically generated images and a method of training patches generation from it. We prove the effectiveness of this data by training a lightweight neural network and show how it performs in both general and documents patches matching. The training was done on the provided dataset in comparison with HPatches training dataset and for the testing, we solve HPatches testing framework tasks and template matching task on two publicly available datasets with various documents pictured on complex backgrounds: MIDV-500 and MIDV-2019.

455-464

Authors: A.V. Shonenkov, D.K. Karachev, M.Y. Novopoltsev, M.S. Potanin, D.V. Dimitrov, A.V. Chertok

We introduce two data augmentation techniques, which, used with a Resnet-BiLSTM-CTC network, significantly reduce Word Error Rate and Character Error Rate beyond best-reported results on handwriting text recognition tasks. We apply a novel augmentation that simulates strikethrough text (HandWritten Blots) and a handwritten text generation method based on printed text (StackMix), which proved to be very effective in handwriting text recognition tasks. StackMix uses weakly-supervised framework to get character boundaries. Because these data augmentation techniques are independent of the network used, they could also be applied to enhance the performance of other networks and approaches to handwriting text recognition. Extensive experiments on ten handwritten text datasets show that HandWritten Blots augmentation and StackMix significantly improve the quality of handwriting text recognition models.

473-478

Authors: V.A. Komarov, A.A. Pavlov

A problem of determining the fiber volume fraction in fiber-reinforced strands of fabric-based laminated composites is considered. As a source of information about the structure of the material, digital micrographs of the ground surface of the cross-sections of the composites are used. Methods and features of the analysis of raster microscopic images of heterogeneous material associated with variable pixel brightness and blurring of the "fiber-binder" boundaries are discussed. To make the image processing less labor-intensive and more objective, a special autoencoder is proposed and built. The study of the structure of a typical structural carbon fiber-reinforced plastic is illustrated by an end-to-end demonstration example. A significant acceleration of the image processing process using the convolutional autoencoder and a good agreement of the results with a careful manual analysis are shown.

483-491

Authors: S.V. Pavlov, A.V. Sokolova, O.I. Khristodulo

In this article, we consider an approach to integrating geoinformation technologies and digital image processing methods aimed at increasing the efficiency of computer technologies in controlling complex distributed systems (CDS) in emergency situations. The analysis is conducted using an example of the activity of oil pipeline transport enterprises, including pipelines, oil pumping stations, administrative, repair and emergency response units. The novelty of the proposed version of integration of geoinformation technologies and digital image processing consists in the application of a set-theoretic approach to a uniform formal description of various types of spatial data, digital images and their interrelationships necessary for constructing three-dimensional models of production facilities of CDS. This has allowed criteria, composition and structure of these 3D models to be defined using different levels of detail of the object models in order to reduce the computational load and speed up the 3D visualization of objects of CDS. These are necessary to support the decision-making when responding to certain types of emergency situations. The advantage of the proposed approach is that it increases the efficiency of decision-making support in countering some emergencies, which, among others, is achieved through the use of three-dimensional visualization of technological and operational (obtained in the event of an accident) information about the production facilities of the CDS, whereas the quality and fullness of the optimized 3D model directly affect the decisions made in response to the emergencies. The developed methods and algorithms have been practically implemented when building 3D geoinformation models.

506-512

Authors: B.S. Mandrikova

A method for analyzing data of complex structure based on combining a wavelet transform and neural networks Autoencoder is proposed. The method allows you to research the data structure, detect abnormal changes of various shapes and durations, and suppress noise. The efficiency of the method is shown on the example of data from a network of neutron monitor stations. Neutron monitor data determine the intensity of secondary cosmic rays and are one of the key factors in space weather. The numerical implementation of the method allows it to be applied on-line, which is of interest in problems of analyzing environmental data and detecting catastrophic events.

184-188

Authors: V.V. Kotlyar, A.A. Kovalev, A.A. Savelyeva

We study a superposition of several identical parallel single-ringed Laguerre-Gaussian beams. Both for the initial plane and for the far field, we show analytically and confirm numerically that if the weighting coefficients of the superposition are real-valued (i.e. all the beams have the same phase, but may have different amplitudes), the net topological charge of the superposition is equal to the topological charge of each constituent beam. It is also shown that a phase delay between the beams allows changing the net topological charge of the superposition. By measuring the net topological charge it is possible to analyze the received optical signal in a data transmission system that utilizes vortex beams, combining spatial multiplexing and orbital angular momentum multiplexing.

196-203

Authors: A.A. Kovalev

In optical communications, it is desirable to know some quantities describing a light field, that are conserved on propagation or resistant to some distortions. Typically, optical vortex beams are characterized by their orbital angular momentum (OAM) and/or topological charge (TC). Here, we study what happens with the OAM of a superposition of two or several optical vortices (with different TCs) when it is distorted by a hard-edge sector aperture. We discover several cases when such perturbation does not violate the OAM of the whole superposition. The first case is when the incident beam consists of two vortices of the same power. The second case is when the aperture half-angle equals an integer number of π divided by the difference between the topological charges. For more than two incident beams, this angle equals an integer number of π divided by the greatest common divisor of all possible differences between the topological charges. For two incident vortex beams with real-valued radial envelopes of the complex amplitudes, the OAM is also conserved when there is a ±(pi)/2 phase delay between the beams. When two beams with the same power pass through a binary radial grating, their total OAM is also conserved.

214-218

Authors: E.M. Pritotskii, A.P. Pritotskaya, M.A. Pankov

Optical characteristics of germanium telluride (GeTe) thin films in multilayer structures are calculated. A model of a multilayer optical coating with a four-level transmission coefficient is developed. Based on the calculated data, thickness values are determined at which the transmittance change is greatest for 1550-nm modulated optical radiation. Experimental samples coated with antireflection GeTe films are synthesized and their optical transmission characteristics are investigated. Combinations of parameters of the multilayer structures for the implementation of contrast transmittance levels are determined. The results of the study represent the implementation of a multilevel artificial optical synapse for neuromorphic processors.

224-231

Authors: V.A. Andreev, A.V. Bourdine, V.A. Burdin, M.V. Dashkov

The paper proposes a method for non-destructive testing of the strength of an optical fiber based on estimates of the energy of nonlinear acoustic emission, based on the use of the tested optical fiber as an acoustic sensor. Models of the processes on which the method is based are presented. Results of experimental studies are presented. Data obtained from testing samples of the optical cables and optical fibers by the proposed method are compared with measurement results for the same samples obtained by the known 2-point method. The error in the estimates of the relative strength of the tested samples of optical fibers is found not to exceed 3.0 %. The results obtained allow us to suggest that the models used are correct and the proposed method shows promise for non-destructive testing of the strength of optical fibers, with the further development of the considered approach showing prospects for the application for non-destructive testing of the strength of optical fibers in a cable, including in-service cables.

239-245

Authors: A.M. Ageev, V.G. Bondarev, V.V. Protsenko

An expedient spectral range of radiation sources for use in the automatic landing system of unmanned aerial vehicles has been substantiated. A method is proposed for synchronizing the photoexposure of the computer vision system and radiation from landmarks for the unambiguous determination of their mutual position. Results of the experimental studies are presented.

252-270

Authors: K.B. Bulatov, E.V. Emelianova, D.V. Tropin, N.S. Skoryukina, Y.S. Chernyshova, A.V. Sheshkus, S.A. Usilin, Z. Ming, J.-C. Burie, M.M. Luqman, V.V. Arlazarov

Identity documents recognition is an important sub-field of document analysis, which deals with tasks of robust document detection, type identification, text fields recognition, as well as identity fraud prevention and document authenticity validation given photos, scans, or video frames of an identity document capture. Significant amount of research has been published on this topic in recent years, however a chief difficulty for such research is scarcity of datasets, due to the subject matter being protected by security requirements. A few datasets of identity documents which are available lack diversity of document types, capturing conditions, or variability of document field values. In this paper, we present a dataset MIDV-2020 which consists of 1000 video clips, 2000 scanned images, and 1000 photos of 1000 unique mock identity documents, each with unique text field values and unique artificially generated faces, with rich annotation. The dataset contains 72409 annotated images in total, making it the largest publicly available identity document dataset to the date of publication. We describe the structure of the dataset, its content and annotations, and present baseline experimental results to serve as a basis for future research. For the task of document location and identification content-independent, feature-based, and semantic segmentation-based methods were evaluated. For the task of document text field recognition, the Tesseract system was evaluated on field and character levels with grouping by field alphabets and document types. For the task of face detection, the performance of Multi Task Cascaded Convolutional Neural Networks-based method was evaluated separately for different types of image input modes. The baseline evaluations show that the existing methods of identity document analysis have a lot of room for improvement given modern challenges. We believe that the proposed dataset will prove invaluable for advancement of the field of document analysis and recognition.

280-297

Authors: P. Musa, I. Purwanto, D.A. Christie, E.P. Wibowo, R. Irawan

Topography is the study of an area on the earth's surface. This term relates to the land's slope or contour, which is the interval of elevation differences between two adjacent and parallel contour lines. Topography generally presents a three-dimensional model of object surface relief and an identification of land or hilly areas based on horizontal coordinates such as latitude and longitude, and vertical position, namely elevation. The topography is essential information that must be provided in the execution of building or road construction based on the ground contour. The problem which is the ground contour which can provide visualization topography as a three-dimensional (3D) condition of the ground contour is not normal (non-linear). Another problem is that the traditional measurement techniques with wheel rotation only measure distances and cannot represent the trajectory of the ground contour in 3D. The proposed in-depth evaluation of orientation estimation results in the topography accuracy level. This methodology consists of several processes; Inertia and orientation of an object, Distance measurement, Terrestrial topocentric – Euclidean transformation, and Topography visualization. This research designed a prototype and proposed a new visualization method of the ground contours to reconstruct a topography map between three algorithms; Direct Cosine Matrix-3D Coordinate, Madgwick-3D Coordinate, and Complementary Filter. The methodology was tested and evaluated intensively by direct observation at three measurement locations with different difficulty levels. As a result, the Direct Cosine Matrix-3D Coordinate is able to visualize the ground contours by reconstructing a topography map much better than other methods.

308-316

Authors: Yu.Kh. Ganeeva, E.V. Myasnikov

The problem of personal identification plays an important role in information security. In recent years, biometric methods of personal identification have become most relevant and promising. The article presents a study of a method for identifying a person from iris images using a neural network approach at the stages of segmentation and a feature representation from the data. A description of a dataset used to implement the segmentation stage using convolutional neural networks is presented and access to the segmentation masks of the entire dataset is provided. A method is proposed for extracting a feature representation of the data using pretrained convolutional neural networks to solve a problem of iris classification. A comparative analysis of methods for extracting iris features, including classical approaches and a neural network approach, has been carried out. A comparative analysis of classification methods is carried out, including classical machine learning algorithms, namely, support vector machines, random forest, and a k-nearest neighbors method. The results of experimental studies have shown the high quality of the classification based on the proposed approach.

326-334

Authors: J. Pamplona, C. Madrigal, J. Herrera-Ramirez

The three-dimensional perception applications have been growing since Light Detection and Ranging devices have become more affordable. On those applications, the navigation and collision avoidance systems stand out for their importance in autonomous vehicles, which are drawing an appreciable amount of attention these days. The on-road object classification task on three-dimensional information is a solid base for an autonomous vehicle perception system, where the analysis of the captured information has some factors that make this task challenging. On these applications, objects are represented only on one side, its shapes are highly variable and occlusions are commonly presented. But the highest challenge comes with the low resolution, which leads to a significant performance dropping on classification methods. While most of the classification architectures tend to get bigger to obtain deeper features, we explore the opposite side contributing to the implementation of low-cost mobile platforms that could use low-resolution detection and ranging devices. In this paper, we propose an approach for on-road objects classification on extremely low-resolution conditions. It uses directly three-dimensional point clouds as sequences on a transformer-convolutional architecture that could be useful on embedded devices. Our proposal shows an accuracy that reaches the 89.74 % tested on objects represented with only 16 points extracted from the Waymo, Lyft's level 5 and Kitti datasets. It reaches a real time implementation (22 Hz) in a single core processor of 2.3 Ghz.