Search

In this paper, the problem of segmentation of halftone images is considered, in which areas of local maxima and minima (extrema) are distinguished with a monotonic change in the brightness of pixels from local extrema to the boundaries of areas. To solve this problem, a mathematical model is proposed and a segmentation algorithm is developed on the basis of counter-wave growing of local extremum regions. The developed algorithm differs from the known segmentation algorithms by using a set of brightness thresholds (by the number of regions), varying by one in each cycle, starting from the values of local extrema, taking into account the increase or decrease in brightness to select adjacent pixels that are attached to the regions formed from these local extrema. The algorithm provides a greater deviation of pixel brightness from the average value within the region compared to known segmentation algorithms. This does not allow evaluating its efficiency using known indicators based on the variance of the brightness within the region. In this regard, estimates of the monotonicity of changes in the brightness of regions are proposed based on a) the shortest distances from each pixel of the region to the corresponding local extremum along the routes determined by the maximum increase (for the region of the local maximum) or decrease (for the region of the local minimum) the brightness of pixels and b) taking into account the number pixels that break the monotony of the segment brightness change. Using these estimates, it is shown that the proposed algorithm provides segmentation of artificial and natural grayscale images with a monotonic change in the brightness of pixels in the areas of local extrema. These properties allow us to consider the developed algorithm as a basis for the selection of texels, spots, low-contrast objects in images.

Objective. The objective of the study is to determine various stable characteristics of images (semi-invariants and invariants) as descriptors necessary for the formation of a feature space of standards intended for recognizing images of different nature belonging to different classes of objects. Methods. The authors propose metrics for evaluating the proximity of the recognized image to a given standard in the space of covariance matrices, based on the obtained descriptors as a methodological basis for constructing image recognition methods. Results. The content of the main stages of selecting descriptors for a given class of objects is developed, taking into account the different illumination of the recognized images. The effectiveness of the results obtained is confirmed by experimental studies related to the solution of the problem of recognition of special images - facies. Conclusions. The definition of stable image descriptors as invariants or semi-invariants to zoom and brightness transformations allows solving the problems of facies classification in conditions of the unstable shooting of recognized images. The images can be rotated and shifted in any way. In general, the proposed approach allows developing an effective image recognition system in the presence of various types of interference on the recognized images.

The article discusses an algorithm for recognizing the contour of fragments of a honeycomb block. The inapplicability of ready-made functions of the OpenCV library is shown. Two proposed algorithms are considered. The direct scanning algorithm finds the extreme white pixels in the binarized image, it works adequately on convex shapes of products, but does not find a contour on concave areas and in cavities of products. To solve this problem, a scanning algorithm using a sliding matrix is proposed, which works correctly on products of any shape., 11 pages, in Russian, 13 figures, ICMTMTE

Technology that allows to localize vertebrae on X-ray images and then determine geometric parameters using the OpenCV computer vision library using a convolutional neural network Darknet YOLO based on regions is proposed.

The radar image formation algorithms for rotating parts of an aerial vehicle in the radar sensor with inverse synthesis of antenna aperture using the monochromatic probing signal are considered. The developed algorithms allow forming radar images of screws oriented horizontally (helicopter main rotor) and vertically (helicopter tail rotor, aircraft propeller). Such radar images will significantly improve the quality of radar recognition of low-flying objects. The estimation of efficiency of algorithms obtained by mathematical modeling is presented.

The results of a study of the application of deep neural convolutional networks to hyperspectral images are presented.

Geometric models for the formation and prediction of multi-angle images are developed. As a means of prediction is proposed to use projective techniques compensate of camera motion. It is shown that the projective motion compensation allows to create an efficient compression techniques of multi-angle images.

Algorithms for fast segmentation of gray scale images based on the block of the wave region growing using quadra-grid pixels are suggested. The algorithms allow to adapt to accessible computational resources and the required accuracy of the segmentation through choosing quadra-grid step.

The technique of the angle optical scan of the flat surfaces with sub-pixel processing of images obtained with the photodetector CMOS matrix scanner is proposed.

Algorithms of sector localization and parameterization of reference points in the wavelet domain for aligning the overlapping images based on the utility of corner coefficients to describe the contour structure in the vicinity of the reference point are proposed. It is shown that these algorithms compared to the methods of SIFT and SURF provide a reduction in the computational complexity of localization and parameterization of reference points and increase the stability of their descriptors at parallax.

The modified algorithms length coding to compress grayscale images without losses that are different from the basic algorithm taking into account the likelihood of the occurrence of pixel values and images that reduce the size of the encoded data.

Developing a method of segmentation and compact multiscale representation of images based on the progressive backward clustering. The method provides an accurate segmentation, multiscale representation and compression of the segmented images, adaptation to a limitation on the time of segmentation.

A method for segmentation of images based on the wave region growing is suggested. In contrast to known methods of segmentation the proposed method allows to divide the regions with smoothing drops of brightness and adapt to the constraints time of segmentation.

A model of reverberation in the synthetic aperture sonar is considered. A assessment of reverberation effect on quality formation of seabed image in synthetic aperture sonar is performed by mathematical simulation with model of reverberant space.

Texture image segmentation method based on estimation of density of contour elements and absorption of small regions, providing an increase in the accuracy in the selection of texture regions in the images to the specification of their boundaries is proposed.

The analysis of quality criteria for evaluating images susceptible subjected to distortions such as «blur» and «defocusing» has been carried out. The criterion for assessing the quality of the recovery process, defined as the degree of similarity of the histogram of the gradients of the reconstructed image with the distribution of brightness gradients obtained a priori, was substantiated.

The results of evaluating the effectiveness of the adaptive two-threshold quantization in comparison with one-threshold quantization for segmentation of gray-scale images using a modified method of separation and merging of areas based on progressive backward clustering are presented.

Spectral method for image search and discovery of the object with the specified geometric characteristics is analyzed. In comparison with the famous search algorithms described algorithm allows to reduce time of search, computational complexity of digital imaging, simplifies the decoding images. The method can be applied in solving problems such as segmentation, image combining, recognition, determining the distance between objects and phenomena, determining the dynamic characteristics of objects.

A method of texture image segmentation based on geometric classification and assessment density of contour elements. The proposed method in comparison with the method based on energy maps, providing a reduction in the error of localization of textural regions by taking into account the geometric characteristics of the elements.

В настоящее время на строительных объектах предпочтение отдается использованию сухих строительных смесей. Сухие смеси позволяют выводить строительные технологии на более высокий уровень, уменьшать потери при производстве и сокращать транспортные расходы. Ассортимент сухих строительных смесей достаточно широк, смеси используют почти на каждом этапе ремонтно-строительных работ. В технологической цепочке производства сухих строительных смесей одним из основных аппаратов является смеситель, способный смешивать порошки, отличающиеся по размеру частиц, их насыпной и истинной плотности и обеспечивать однородность получаемой смеси. Достижение однородности состава конечного продукта зависит от режимных параметров работы используемого оборудования. Целью настоящей работы явилось исследование способов повышения качества сухих смесей, получаемых в вибрационных смесителях. Для этого проведены экспериментальные исследования на установке, представляющей собой прототип промышленного вибрационного смесителя. В ходе эксперимента исследовали влияние режимов работы смесительного аппарата на качество получаемой бинарной смеси. Чтобы определить распределение содержания ключевого компонента по высоте миксера, использовали фотофиксацию процесса и средства цифровой обработки изображений. Анализ полученных фотографий проводили при помощи стандартных функций пакета Image Processing Toolbox, который является частью программной среды MATLAB. С целью прогнозирования эксперимента предложили его математическую модель. Процесс получения сухих смесей в вибрационном аппарате описали математической моделью на основе теории цепей Маркова, которая реализует стохастический подход к изучению процесса и позволяет определять оптимальное время смешивания компонентов. Рабочая камера миксера представлена массивом ячеек идеального смешения, через которые осуществляется движение компонентов смеси согласно матрице переходных вероятностей. Текущее состояние смеси характеризуется вектор-столбцом. Эволюция состояния смеси описывается рекуррентным матричным равенством. Однородность смеси оценивается среднеквадратичным отклонением от идеального распределения компонентов смеси по ячейкам. На основе представленной модели приведены компьютерные методы расчета процесса получения сухих строительных смесей в вибрационных смесителях, реализованные в матричной среде пакета MATLAB., Nowadays dry mortars are widely used at construction sites. As a result, construction technologies are enhanced, which reduces production losses and transportation costs. The range of dry mortars is quite wide; they are used at almost every stage of repair and construction work. In the technological chain of dry mortars production, the main apparatus is the mixer that can mix powders that differ in particle size, bulk and true density to ensure the uniformity of the resulting mixture. Achieving uniformity of the final product composition depends on the operating parameters of the equipment used. This research is aimed at the improvement of the quality of dry mixtures obtained in vibrating mixers. For this purpose the authors have carried out experimental studies on the prototype of an industrial vibration mixer. In the course of the experiment, the influence of the operating modes of the mixing apparatus on the quality of the resulting binary mixture was investigated. In order to determine the distribution of the content of the key components over the height of the mixer, photographic recording of the process and digital image processing were used. The analysis of the obtained photographs was carried out using the standard functions of the Image Processing Toolbox package, which is part of the MATLAB software. For the accuracy of the experiment its mathematical model has been proposed. The process of obtaining dry mortars in a vibrating apparatus is described by a mathematical model based on the theory of Markov chains, which implements a stochastic approach to studying the process and makes it possible to determine the optimal mixing time of the components. The working chamber of the mixer is represented by an array of ideal mixing cells through which the mixture components move according to the matrix of transition probabilities. The current state of the mixture is characterized by a column vector. The evolution of the state of a mixture is described by recurrent matrix equality. The homogeneity of the mixture is estimated by the root-mean-square deviation from the ideal distribution of the mixture components over the cells. Based on the developed model, the authors propose computer methods for calculating the process of obtaining dry mortars in vibrating mixers using MATLAB package., ВЕСТНИК ПОВОЛЖСКОГО ГОСУДАРСТВЕННОГО ТЕХНОЛОГИЧЕСКОГО УНИВЕРСИТЕТА. СЕРИЯ: МАТЕРИАЛЫ. КОНСТРУКЦИИ. ТЕХНОЛОГИИ, Выпуск 1 (17) 2021

Операция выравнивания гистограмм (НЕ – Histogram equalization) часто используется для увеличения качества изображения, увеличения контраста. В рамках этого раздела рассматривается понятие гистограммы, последовательность действий для ее выравнивания, практические примеры работы этого метода. Следует отметить, что обработку изображений целесообразно производить непосредственно в сжатом формате (JPEG) без распаковки. Обработка изображений в сжатом виде обеспечивает более значимую эффективность и требует меньших вычислительных затрат по сравнению с методом, использующим распаковку изображения из сжатого формата, обработку изображения с дальнейшим сжатием его. Для обработки изображений мы будем использовать библиотеку OpenCV вместе с доступной версией языка C/С++, C# или Python. Для сравнительного анализа скорости обработки изображений разными методами более предпочтителен вариант с использованием языка C/С++, кроме того, этот вариант даст возможность организовать обработку изображения несколькими потоками, что значительно ускорит обработку. Вариант использования интерпретирующей программной среды Python, дополненной методами, программами и библиотеками, широко представленными в информационной сети, следует рассматривать как экспериментальный полигон для отработки тех или иных методов обработки данных., The operation of histogram equalization (NOT – histogram equalization) is often used to increase the image quality, increase the contrast. Within the framework of this section, the concept of a histogram, procedures for its alignment, practical examples of how this method works. It should be noted that image processing should be done directly in compressed format (JPEG). Compressed image processing provides more significant efficiency and requires less computational costs compared to the method using decompression of an image from a compressed format, image processing with its further compression. For image processing, we will use the OpenCV library along with the available version of C / C++, C# or Python. For a comparative analysis of the speed of image processing by diff erent methods, the option using the C / C++ language is more preferable, in addition, this option will make it possible to organize image processing in several threads, which will signifi cantly speed up the processing. The option of using the Python interpreting software environment, supplemented by methods, programs and libraries widely represented in the information network, should be considered as an experimental testing ground for working out certain methods of data processing., Промышленные АСУ и контроллеры, Выпуск 12 2021

The urgency of the task of developing tools for neural network analysis of biometric parameters for recognizing the personality and emotions of students of the distance learning system has been substantiated. The necessity of formalizing the architectural solutions used in the creation of software for neural network analysis of biometric parameters is shown. As a result of the research carried out in terms of the UML modeling language, the architecture of the neural network analyzer of biometric parameters has been developed. Diagrams of options for using the neural network analyzer have been developed both for recognizing the personality of a student when entering the system, and for recognizing the personality and emotions of a student in the process of his interaction with the distance learning system. Also, based on the developed use case diagrams, a structural diagram of the analyzer is built. The necessity of including subsystems for determining the functional parameters of the analyzer, registration of biometric parameters, neural network analysis of registered biometric parameters, personality recognition and emotion recognition is substantiated. An original feature of the proposed architectural solutions is the introduction into the neural network analysis subsystem of an integrated analysis module designed to summarize the results of neural network analysis separately for each of the biometric parameters. A rule for making an integrated decision has been developed, taking into account the results of a neural network analysis of each of the registered biometric parameters and the corresponding weight coefficients determined by expert evaluation. The introduction of the integrated analysis module makes it possible to increase the accuracy of recognition of emotions and personality of a student, since the final classification is realized through a generalized assessment of several guaranteed significant biometric parameters. In addition, the use of this module makes it possible to increase the reliability of the neural network analyzer in case of difficulties associated with the registration of a particular biometric parameter. It has been established that the decision-making rule can be improved by using one or more neural networks in the integrated analysis module, designed to generalize the results of the neural network analysis of all registered biometric parameters. It is proposed to correlate the directions of further research with the development of appropriate neural network solutions., Обоснована актуальность задачи разработки инструментальных средств нейросетевого анализа биометрических параметров для распознавания личности и эмоций слушателей системы дистанционного обучения. Показана необходимость формализации архитектурных решений, используемых при создании программных средств, предназначенных для нейросетевого анализа биометрических параметров. В результате проведенных исследований в терминах языка моделирования UML разработана архитектура нейросетевого анализатора биометрических параметров. Разработаны диаграммы вариантов использования нейросетевого анализатора как при распознавании личности студента при входе в систему, так и при распознавании личности и эмоций студента в процессе его взаимодействия с системой дистанционного обучения. Также, базируясь на разработанных диаграммах вариантов использования, построена структурная схема анализатора. Обоснована необходимость включения в состав анализатора подсистем определения функциональных параметров анализатора, регистрации биометрических параметров, нейросетевого анализа зарегистрированных биометрических параметров, распознавания личности и распознавания эмоций. Оригинальной особенностью предложенных архитектурных решений является внедрение в подсистему нейросетевого анализа модуля интегрированного анализа, предназначенного для обобщения результатов нейросетевого анализа отдельно каждого из биометрических параметров. Разработано правило принятия интегрированного решения, учитывающее результаты нейросетевого анализа каждого из регистрируемых биометрических параметров и соответствующие им весовые коэффициенты, определенные путем экспертного оценивания. Внедрение модуля интегрированного анализа позволяет повысить точность распознавания эмоций и личности студента, поскольку окончательная классификация реализуется за счет обобщенной оценки нескольких гарантированно значимых биометрических параметров. Кроме этого, использование данного модуля позволяет повысить надежность нейросетевого анализатора в случае трудностей, связанных с регистрацией того или иного биометрического параметра. Установлено, что правило принятия решения возможно усовершенствовать за счет использования в модуле интегрированного анализа одной или нескольких нейронных сетей, предназначенных для обобщения результатов нейросетевого анализа всех регистрируемых биометрических параметров. Предложено соотнести направления дальнейших исследований с разработкой соответствующих нейросетевых решений.

The purpose of this article is to introduce clustering algorithms and apply clustering methods to solve the problem of segmentation of medical images.

Properties of a popular measure of comparing a digital image with a reference – the index of structural similarity, called SSIM in the literature – are explored. It is proved that the SSIM and its derivative functions are not metrics. Variants of the index modification are described. It is shown that measures similar to this index evaluate not quality of images, but their similarity by fragments. Additionally, it is shown that the averaged expert assessments called MOS are very subjective and cannot exact correlate with numerical estimates of similarity of the compared images. To get the SSIM index, a matrix of local estimates is calculated. Each evaluation determines similarity of two analyzed pixels with the same coordinates taking into account neighboring pixels. Usually, the average arithmetic value of the obtained matrix is used as the SSIM index. Instead, to improve the SSIM index, it is proposed to use the scale parameter of the Weibull distribution, which approximates the histogram of the local index estimates. On a set of images from the public database TID2013, it is shown that the proposed parameter is a more accurate characteristic of image similarity than the mean of local estimates.

Transformation of a digital radar image with a wide dynamic range of luminance values (up to 216 tones) for visualization on a standard monitor is considered. The monitor has 28 gray value range or any of three basic colors in the RGB system. Linear quantization of the luminance values of the original image generates a nearly black image with small light blobs, since most of the original values are less than 255–512 and are reduced by 256 times. To solve the problem, it is proposed to use a nonlinear logarithmic transformation with a parameter calculated from the original images. On real data received from the TerraSAR-X satellite in the geoTIFF format, comparative studies with other compression algorithms for a wide dynamic range of image brightness are performed. It is shown that the proposed solution allows creating a visually better image in comparison with the known algorithms.

Application of nonparametric transformations for construction of local image features descriptors is considered. An approach based on scalable Bresenham circles for multiring nonparametric transformations is proposed. A problem of finding characteristic points (corners) in images is solved using two rings of small radius. Then, considering rings of greater radius a final decision is made concerning local features under investigation. Testing results of proposed local descriptors have demonstrated their competitiveness with known descriptors.

The influence of two rectangular contours with their various configurations is considered. The formulas for calculation of mutual induction are received and based on them the dependences of the coefficient of mutual induction of the distance between the contours with various shapes are built. The connection of the coefficients of mutual induction for a variety contours is installed.

Algorithms coordinate value of the aircraft relative to the earth's surface by means of a digital camera or vision systems, as well as the initial alignment algorithm for the navigation system are presented.

A new method of clustering of grayscale, color and multispectral images is presented. It is based on conditional optimization of the objective function consisting of the classic fuzzy functional criterion and the penalty function of Gibbs type, which controls local smoothness of the solution. The method provides more smooth solutions that are essentially more precise in comparison with fuzzy c-means results in the case of noisy images.

An approach to developing informative descriptors for solving the problem of statistical recognition of cryptographic generators is proposed. The descriptors developed by this approach are given. Theoretical results are illustrated by the computer experiments.

Obtaining images from γ-sources is a problem when treating radioactive ma-terials. There are three main methods for constructing gamma-imagers: space scanning with a collimator, using coding masks and the Compton camera. Also, a combination of them may be realized.

The paper proposes a method for recognition of three-dimensional objects using deep machine learning. In computing systems, objects are often represented by 3D models as a set of polygons or surfaces describing a geometric shape. Search for relevant objects by recognizing on the basis of such data without compression is ineffective. In addition, when recognizing on the basis of pairwise comparison of objects with each other, it is often difficult to systematize the results. The proposed recognition method is aimed at solving these problems. In recognition, spectral descriptors are applied using characteristics that describe various physical processes on the surface. These descriptors use the spectral decomposition of a discrete analogue of the Laplace-Beltrami operator for objects whose surface is approximated by a triangular grid. Each of the presented objects is represented by three descriptors. A method of compressing information about the shape of an object represented by descriptors using the entered spectral distribution maps is proposed. The peculiarity of this compression method is that when it is used it: is possible to compare objects of different levels of detail, the recognition process is accelerated, and the important properties of resistance to noise and invariance to various form transformations possessed by spectral descriptors are preserved. Object recognition is then performed using deep machine learning, which uses a convolutional neural network with several channels. The input data for each channel of the neural network are maps of spectral distributions. Recognition is performed by computing in a pre-trained neural network and then determining the class to which the object belongs. A series of computational experiments using various configurations of spectral descriptors was carried out. Experimental results demonstrate high recognition accuracy for three-dimensional objects with various shape transformations.

Three versions of a histogram algorithm for tracking objects on video sequences made by an unstable camera are presented. Local color 1D-histograms of pixels and local color 2D-histograms of pairs of adjacent pixels are used in all versions as region features. The histograms are compared by the Bhattacharia criterion. А parallel computing platform CUDA, developed to program GPUs, allows creation of real time or near-real time program realizations of the offered versions. Results of comparison of the versions with the known mean shift algorithm and correlation type algorithms are also presented. It is shown by experiments that the versions are more accurate and reliable than the mean shift algorithm, which estimates similarity of linear approximations of local histograms, and more robust with respect to the video quality than the correlation algorithms.

The method of non-linear amplitude suppression of radar clutter from a sea surface in a noncoherent band path of the receiver is considered. The model of the most adequate nongaussian a radar clutter from the rough sea surface which envelope is based on K-distribution of density of probability is used. In model of noises it is considered internal gaussian receiver noise. For various values of parameters of model of clutter amplitude characteristics of a non-linear element and dependence of coefficient of suppression on the relation a clutter/noise are under construction.

Methods of panoramic radiomonitoring and developed on their basis algorithms for joint detection and direction finding in a wide band of frequencies are presented. The block diagrams and panoramas illustrating the results of the algorithms are given.

An algorithm for manipulating pixels of image is proposed. It is used to modify the method of identifying several dynamic objects.

The article presents the results of an analysis of the features functioning of Hamming discrete neural network, which can not recognize input black-and-white images that are at the same minimum distance from two, three or four reference images. The shortcomings of neural networks, the Hamming distance and solving this problem for images located on the boundaries of two or three classes of images are analyzed. A modification of the Hamming neural network, recognizing images on the boundaries of several classes, is proposed., В статье изложены результаты анализа особенностей функционирования дискретной нейронной сети Хемминга, которая не может распознавать входные черно-белые изображения, находящиеся на одинаковом минимальном расстоянии от двух, трех или четырех эталонных изображений. Проанализированы недостатки нейронных сетей, расстояние Хемминга и решающих эту задачу для изображений, находящихся на границах двух или трех классов изображений. Предложены модификация нейронной сети Хемминга, распознающей изображения на границах нескольких классов.

The method of non-linear amplitude suppression of radar clutter from a sea surface in the coherent band path of the receiver is considered. The model of the most adequate nongaussian radar clutter from the rough sea surface which envelope is based on K-distribution of density of probability is used. The mathematical model of clutter is considered. For various values of parameters of model of clutter amplitude characteristics of a non-linear element and dependence of coefficient of suppression on the relation a clutter/noise are under construction.

Problems of image comparison of visual objects in systems of computer vision are considered. The measures of similarity which takes into account distortion and occurrence of the false component in the structural description are suggested. The analysis of properties of methods is implemented, results of computer modeling for real images are described., Рассматриваются вопросы сопоставления изображений визуальных объектов в системах компьютерного зрения. Предложены меры сходства, учитывающие искажение и появление ложных компонент в структурном описании. Проведен анализ свойств методов. Описаны результаты компьютерного моделирования для реальных изображений., Розглянуто питання зіставлення зображень візуальних об’єктів у системах комп’ютерного зору. Запропоновано міри схожості, які враховують викривлення та появу хибних компонент у структурному описі. Проведено аналіз властивостей методів, описано результати комп’ютерного моделювання для реальних зображень.

The paper considers basic model parameters of a static imaging Fourier spectrometer for Earth remote sensing from space. Design of the imaging spectrometer model has been developed and images have been obtained in the paper. The images have made it possible to determine such parameters as ratio of signal/ noise, frequency/contrast characteristic and spectral coincidence of reconstructed images.

A mask-phase method of localization of straight lines in an image is proposed. It is based on a preliminary quantization of line segments with respect to orientation by using a library of binary masks and rectification of straight lines by using oriented Hough transform. The method allows improving the speed and quality of straight lines localization if there are changes of brightness, contrast and image noise.

This article presents an algorithm for restoring broken pixels, which can appear in graphic files with statistical gaps. The suggested algorithm is based on the method of hierarchical analysis of the decision support theory. The choice of the broken pixel color depends on the nearest neighbors and their next neighbors. Three parameters inherent in each nearest neighbor are analysed. Firstly, it is the number of neighbors, which have the same color as the given nearest neighbor. Secondly, it is a deviation of the given pixel from the average value of its neighbors. Thirdly, it is a difference between the pixels being on the opposite sides of the broken pixel. Based on these three criteria, for each nearest neighbor of the broken pixel, a weight coefficient is defined. A hierarchical two-level tree for making decision is constructed. As a color of the broken pixel, its neighbor color with the maximum weight is chosen.A computer experiment to determine the effectiveness of the proposed method is conducted. The effectiveness of the proposed method was determined by comparing the similarity degree of the broken and restored images to the source image. To compare images Minkowski metric was used. To conduct experiments photographic and artificial images were used. The paper investigates a dependence of the proposed algorithm efficiency on the broken image amount. It was found out that the proposed algorithm has the advantage over the known algorithms for restoring broken pixels near the sharp edges. An image restored by our method has more sharply defined edges as compared to what the smoothing filters provide. The proposed method can be iteratively applied. As the experiments have shown, the first five iterations provide image enhancement.The proposed method together with the algorithms for detecting the broken pixels can be used to design filters of noisy images. The method efficiency enhancement can be achieved in case of taking into account the large number of neighbors in making decision. This method can be also developed using the increased number of criteria when making decisions.

An algorithm for the extraction of image contours is presented. The algorithm is based on combinatorial methods of clustering the oriented gradient of an image. It allows to estimate the position and the slope angles of the contours with a sufficient accuracy, and to obtain their vector representation by the broken lines or pieces of smooth curves, which is convenient for analyzing the shape of objects.

This article describes the use of clustering for face recognition image. Clustering was performed using a recurrent neural network used at two stages of the recognition process. The algorithm includes the recognition process itself perform clustering pixel brightness image, calculating image information close proximity and clustering measures to in order to obtain the cluster containing the original similar images.

This article describes the use of clustering for face recognition image. Clustering was performed using a recurrent neural network used at two stages of the recognition process. The algorithm includes the recognition process itself perform clustering pixel brightness image, calculating image information close proximity and clustering measures to in order to obtain the cluster containing the original similar images.

Operation principles and application of the aircraft thermal vision scanners are considered. Their workability is described. The operation diagram of the future thermal vision device with adaptive scanning is presented. The calculation methods of the flight altitude taking into account maximum detection range of the object thermal vision emission are presented.

This paper describes a realization of method of Human-Computer interaction based on using video-stream. Described method applies Viola-Jones algorithm and Hidden Markov models for gestures capturing and recognition.

A method for low-contrast image processing and analysis in the space of ellipsometric Stokes parameters, which can be applied to improve the quality of one-parameter and multi-parameter images is proposed in the paper. Under the method, each pixel of the analyzed image is compared with four virtual Stokes parameters, which allows naturally implement color RGB coding of results without using the pseudocolor coding procedure. The proposed results interpretation method using RGB coding based on the parameters improves the low-contrast image analysis reliability.Experiments have shown that the most appropriate method for multispectral ensemble compression is a singular value decomposition method. Images, corresponding to the first three maximum singular values are taken as three main (“own”) images. The synthesis of elliptic characteristics based on their “own” images provides a higher degree of detail of the resulting color elliptic RGB coding compared with direct RGB coding of three “own” images.Presented information technology allows to carry out low-contrast image processing using the described methods and is designed to increase the sensitivity of visual analysis and selection of objects of potential interest., Исследованы информационные возможности нового метода для обработки и выделения объектов интереса на слабоконтрастных изображениях, основанного на проведении виртуальных аналогий с методами оптических физических измерений и рассмотрении в качестве информативных параметров пространства эллипсометрических параметров Стокса. Представлены результаты проверки работоспособности метода на изображениях различной физической природы (геофизическое поле, космоснимок, томограмма)., Досліджено інформаційні можливості нового методу для обробки та виділення об’єктів інтересу на слабоконтрастних зображеннях, який базується на проведенні віртуальних аналогій з методами фізичних вимірів і розгляду в якості інформативних параметрів простору еліпсометричних параметрів Стокса. Приведено результати перевірки працездатності методу на зображеннях різної фізичної природи (геофізичне поле, космознімок, томограма).

It is considered the determination of impact force and frequency, shock acceleration, impact torques while bottom slamming for acute vessels with bow bulb, when slamming characteristics are defined by hydro-dynamics bulb but not the hull impacts., Рассмотрено определение частоты и силы ударов, а также ударных ускорений и ударных изгибающих моментов при днищевом слеминге для острого судна с носовым бульбом, когда характеристики слеминга определяются гидродинамическими ударами бульба, а не корпуса., Розглянуто визначення частоти і сили ударів, а також ударних прискорень та ударних згинальних моментів в умовах днищевого слемінгу для гострого судна з носовим бульбом, коли характеристики слемінгу визначаються, насамперед, гідродинамічними ударами бульба, а не корпусу.

This article is devoted to recognition of musical instruments sound in a complex signal based on their spectral properties. The recognition is performed by a neural network using mel-frequencies cepstral coefficients as features.