Новости

On this page, talks in the format of posters are posted. They can be viewed during the whole time of the Conference. To send your question to the authors, please follow to Miro and write your question in the comments. You can also use the feedback form on our website. Authors will answer questions using Miro comments or via email. To get manual go to Instructions (eng / rus).

Vladislav Myasnikov and Alexander Verichev
Image Inpainting as a Quadratic Programming Task
Digital image inpainting is an image restoration technique used to replace missing, damaged, or deteriorated parts of images. Classical textual and structural, as well as deep neural network learning-based methods have been proposed for inpainting digital images originating from various sources. In this paper we propose a method that falls within the realm of classical methods. The inpainting task is formulated as a quadratic programming problem, and two solution methods, unsupervised and supervised, are described. The proposed methods are experimentally evaluated on the various classes of test images. We analyze and discuss the results and identify possible future research directions.

Vladimir Chernov
Discrete orthogonal transforms with self-similar basis functions
New bases of discrete orthogonal transforms associated with some recursive processes and possessing the property of self-similarity are introduced and investigated. Sufficient conditions for the orthogonality of the system of basis functions are proved. Fast transformation algorithms are synthesized for transformations with introduced bases. The connection of the bases under consideration with the analytical properties of the Dirichlet generating series is discussed.

Artem Mukhin, Igor Kilbas, Rustam Paringer and Nataly Ilyasova
Application of the gradient descent for data balancing in diagnostic image analysis problems
The article proposes an algorithm for data balancing based on gradient descent. The proposed algorithm is able to partially mitigate the influence of the data imbalance problem which is commonly seen in the tasks of diagnostic image analysis. The authors have investigated the influence of the proposed algorithm on the accuracy of a fully convolutional neural network. The neural network was trained on unbalanced data as well as on the balanced by the algorithm. Recommendations on how to use the proposed algorithm are also formulated.

Ekaterina Galitskaya and Viktor Krasheninnikov
Ways to increase the probability of correct recognition of noisy speech commands by their cross-correlation portraits
Currently, the field of application of voice information-control systems is being intensively expanded, for which recognition of speech commands (SC) is necessary. This recognition is very difficult in the presence of intense acoustic noise. We consider a method for recognizing noisy SCs by cross-correlation portraits (CCP), which is used for speaker-dependent recognition from a limited vocabulary of commands. In this method, SCs are converted to CCPs, which are special images. The probability of correct recognition directly depends on the choice of command standards. The standards should accurately reflect the entire class of commands, for which the library of standards is optimized. The standards are stored as CCPs. Recognized SC is converted into CCP and the closest portrait is found from the set of portraits of standards. In this case, a sufficiently accurate coincidence of the portraits of the standard and the recognizable SC is required. For this, two methods are proposed: phonemic alignment and variation of the boundaries of SC, given that its boundaries can be estimated ahead or delayed. The experiments showed that the proposed modernization of the algorithm significantly increases the probability of correct recognition.

Nikita Andriyanov and Danila Andriyanov
The importance of data augmentation in machine learning for image processing tasks in the face of data scarcity
The article presents the results of alphabet character recognition by various neural networks in the limited conditions of the source data and with a number of simple augmentations. Furthermore the dependences were obtained for a serial neural network with back propagation of error. The simplest transformations were used for data augmentation. Augmentation process includes the slope of the characters (italics), changing the colour of the characters (from black to red), as well as distortion of the reference images with white Gaussian noise at a signal-to-noise ratio q = 1 ... 10. It is shown that the best results of recognition of the Russian alphabet characters are provided by a network for which all the augmentation methods discussed in this work were used. A study was also conducted of the dependence of recognition accuracy on the signal-to-noise ratio in all trained neural networks.

Nikita Andriyanov
Using neural networks to identify parameters of autoregression model with multiple roots of characteristic equations
To determine the parameters of autoregressive models with multiple roots of characteristic equations, neural networks were trained. Special cases for first and second order models are considered. Efficiency studies were conducted for neural networks with back propagation of error, cascade type networks and generalized regression networks. The results are compared with an identification algorithm based on correlation functions. The proposed algorithms are implemented both for one-dimensional random sequences based on such models, and for two-dimensional random fields. In addition, the parameters were identified for fitting models to real images.

Nikita Andriyanov and Danila Andriyanov
Modeling and processing of SAR images
The article is devoted to the method of simulating radar images based on harmonic analysis. The possibilities of forming small and distributed objects in the coordinates of inclined and track ranges are considered. Moreover, for a number of reference objects, a detection algorithm based on the Neyman-Pearson criterion was implemented and investigated, and an algorithm for recognizing reference targets was proposed.

Irina Gndoyan, Alexey Petraeyvsky, Victor Fedoseev and Maria Denisenko
Determination of quantitative and qualitative indicators of hemomicrocirculation of the anterior eye segment according to the results of non-invasive application fluorescein angiography
Non-invasive application fluorescein angiography of the anterior eye segment provides valuable information about the capillaries and venules hemomicrocirculation of the anterior eye segment in inflammatory and dystrophic ocular diseases. This information includes some quantitative (the number of functioning peripheral and prelimbal venules, the time vessels of the complete filling with the dye) and qualitative (the presence and great of extravasal hyperfluorescence) indicators. However, these indicators are usually assessed visually from digital images. This approach leads to inaccuracies and loss of the important information. In addition, some indicators (for example, the total area of the microvascularization) cannot be obtained "manually." Thus, in this paper, we propose a digital image processing technique aimed to evaluate such parameters. It includes blood vessels segmentation, skeletonization, vessel width estimation, and calculation of the resulting indicators

Vadim Turlapov, Tamara Utesheva and Konstantin Pukhky
The task of detecting the boundaries of hyperspectral image objects
The problem of boundary detection by the Canny (John F. Canny) algorithm is investigated as a complementary tool for the analysis, segmentation and classification of the hyperspectral (HSI) and multisensor images objects. The possibilities of various measures of the distance between the k-dimensional vectors of signatures in the detection of classes and states of HSI objects are investigated: the angular distance (in the form of the cosine of the angle); Pearson correlation coefficient; Euclidean norms. First of all, the possibilities were analyzed in a situation where the object of interest is determined by the features that appear in part of the HSI channels. Based on the feature vector, an object boundary is detected. Then the object inside the boundary is examined in another part of the channels (or in all channels) by the histogram of the corresponding metric or by the values in the individual channels. An adaptation of the John F. Canny algorithm has been implemented to detect the boundaries of the region of interest as a tool for the study and classification of HSI objects, which creates new opportunities for analysis. The angular distance is determined as the leading scalar metric for detecting boundaries. Values of standard deviations, an average of the signature, Euclidean norms of signatures are used as features of the second level classification. The references of the contoured objects can be used as references of the state of the object for comparative studies, and for further unmixing in units of the library reference objects.

V.O. Sokolov
75 years of prof. V.A. Fursov
Article briefly tells about the life and scientific activity of Professor Vladimir A. Fursov. Professor Vladimir Fursov is the Head of the Department of Supercomputers and General Informatics of the Samara National Research University named after academician S.P. Korolev (Samara University) and the leading researcher of the Image Processing Systems Institute of the Russian Academy of Sciences (IPSI RAS) - the branch of the Federal Scientific Research Center “Crystallography and Photonics” of the Russian Academy of Sciences. Professor Vladimir Fursov is a well-known specialist in the field of image processing and pattern recognition. The article analyzes his recent contribution to the development of the theory of identification, methods of images processing and recognition and supercomputing technologies.

Aleksandr Borodinov and Vladislav Myasnikov
Analysis of user tracks on public transport
The paper considers the problem of matching GPS tracks to a road network. We presented a map-matching algorithm based on dynamic programming. We collected the tracks of movement around the city of several users on personal vehicles with various trip types to test the proposed algorithm. The data collected after matching to the road network can be used to further identify user preferences and to build a transport recommender system.

Ivan Kholopov and Igor Kudinov
Scene-based non-uniformity fixed pattern noise correction algorithm for infrared video sequences
An algorithm for a fixed pattern noise correction for infrared sensors based on the analysis of the video sequence of a static or dynamic scene observed by the camera is considered. It is shown that on the assumption of the additive nature of the fixed pattern noise, the frame-to-frame accumulation of such noise by analogy with the radar problem of detecting a signal against a correlated clutter can successfully compensate for it with a video sequence of more than 500 frames. During experiments with the Xenics Bobcat 640 short-wave infrared and Xenics Goby 384 long-wave infrared cameras it was demonstrated that in contrast to the well-known non-uniformity correction algorithm for a single frame, typical for it halo artifacts near extended scene objects are not observed in the resulting image, when fixed pattern noise is estimated from the results of accumulation over a set of frames.