• A
  • A
  • A
  • ABC
  • ABC
  • ABC
  • А
  • А
  • А
  • А
  • А
Regular version of the site
Of all publications in the section: 22
Sort:
by name
by year
Article
Nikitin M., Konushin V., Konushin A. Computer Optics. 2019. Vol. 43. No. 4. P. 618-626.

Modern biometric systems based on face recognition demonstrate high recognition quality, but they are vulnerable to face presentation attacks, such as photo or replay attack. Existing face anti-spoofing methods are mostly based on texture analysis and due to lack of training data either use hand-crafted features or fine-tuned pretrained deep models. In this paper we present a novel CNN-based approach for face anti-spoofing, based on joint analysis of the presence of a spoofing medium and eye blinking. For training our classifiers we propose the procedure of synthetic data generation which allows us to train powerful deep models from scratch. Experimental analysis on the challenging datasets (CASIA-FASD, NUUA Imposter) shows that our method can obtain state-of-the-art results.

Added: Oct 31, 2019
Article
Savchenko A. Computer Optics. 2013. Vol. 37. No. 2. P. 254-262.

The usage of the probabilistic neural network with homogeneity testing is proposed in image recognition problem. This decision is shown to be optimal in Bayesian terms if the task is formulated as a statistical testing for homogeneity of query and model images' feature sets. The problem of the lack of computing efficiency with many classes and large dimensions of feature set is discovered. The possibility of its overcoming in the case of discrete features is explored by synthesizing the novel recognition criterion with the comparison of the histograms of query and model images. It is shown that a particular case of this criterion is the nearest neighbor rule with popular measures of similarity, namely, chi-square distance and Jensen-Shannon divergence. The results of experimental research in a problem of face recognition with widely used databases (AT&T, JAFFE) are presented. The proposed approach is demonstrated to achieve better recognition accuracy in comparison with conventional solution with reduction the recognition task to the statistical classification.

Added: Jul 1, 2013
Article
Konushin A., Nikitin M., Konushin V. Computer Optics. 2017.

Computer Optics

Added: Feb 7, 2018
Article
Umnov A., Krylov A. S. Computer Optics. 2016. Vol. 40. No. 6. P. 895-903.

In this paper we suggest an algorithm for ringing suppression based on a sparse  representation method. As one of its steps, the suggested method includes image deblurring  based on the Wiener-Hunt deconvolution algorithm. The ringing suppression algorithm uses  the signals' mutual coherence and sparsities analysis when dealing with the ringing effect  based on the sparse representation method. We also analyze the mutual coherence and  sparsities for blurred images and images with white Gaussian noise.

Added: Oct 21, 2017
Article
Savchenko A. Computer Optics. 2012. Vol. 36. No. 1. P. 116-123.

The problem of the choice of algorithms parameters in automatic image recognition is put and solved by ensemble classifiers construction using the maximum posterior probability principle. The new criterion of parameters choice is strictly synthesized for Kullback-Leibler information discrimination and modern SIFT (Scale-Invariant Feature Transform) method of object recognition. The program and results of experimental research in a problem of face recognition with widely used databases (Yale, AT&T) are presented. It is shown that the proposed criterion allows to achieve recognition accuracy equal to the algorithm with the best parameters set, and not only for Kullback-Leibler information discrimination, but also for other popular distances (Euclidean metric, Kullback information divergence).

Added: Feb 4, 2013
Article
Pham Cong T., Копылов А. Computer Optics. 2018. P. 1-8.

We consider here image denoising procedures, based on computationally effective tree-serial parametric dynamic programming procedures, different representations of an image lattice by the set of acyclic graphs and non-convex regularization of a new type which allows to flexibly set a priori preferences. Experimental results in image denoising, as well as comparison with related methods, are provided. A new extended version of multi quadratic dynamic programming procedures for image denoising, proposed here, shows an improved accuracy for images of a different type.

Added: Jul 21, 2018
Article
Евсютин О. О., Шелупанов А. А., Мещеряков Р. В. и др. Компьютерная оптика. 2017. Т. 41. № 3. С. 412-421.

In the paper, we consider a particular direction of digital steganography — information embedding into the compressed JPEG images. A scheme of information embedding based on procedures of DCT-coefficients replacement is introduced. Variants of the scheme algorithmic implementation are offered and investigated. A genetic algorithm is used for the improvement of the embedding quality. The main result of the investigation is a steganographic algorithm of information embedding into the compressed JPEG images. This algorithm utilizes an unstable region of embedding at the level of one block of DCT-coefficients. The choice of an optimum region of embedding is performed by the genetic algorithm.

Added: Sep 4, 2019
Article
Козачок А. В., Копылов С. А., Мещеряков Р. В. и др. Компьютерная оптика. 2017. Т. 41. № 5. С. 743-755.

The development of an Internet of things concept has led to an essential increase in the amount of data processed via the Internet. Multimedia data constitute a significant proportion of this information. This type of data often contains user’s personal information or copyright protected data. The issue of copyright protection of digital imagery has remained topical for the last decades. Traditional information protection tools cannot provide the required level of image protection from possible threats due to specific features of format representation. This article contains a comparative analysis of published research papers concerned with the robust image hashing as one of possible methods of copyright protection of digital imagery. It also includes a classification of robust image hashing methods, discussing their advantages and drawbacks, and their major characteristics. At the end of the article some directions of further research are outlined.

Added: Sep 5, 2019
Article
Савченко А. В. Компьютерная оптика. 2012. Т. 36. № 1. С. 117-124.

The problem of the choice of algorithms parameters in automatic image recognition is put and solved by ensemble classifiers construction using the maximum posterior probability principle. The new criterion of parameters choice is strictly synthesized for Kullback-Leibler information discrimination and modern SIFT (Scale-Invariant Feature Transform) method of object recognition. The program and results of experimental research in a problem of face recognition with widely used databases (Yale, AT&T) are presented. It is shown that the proposed criterion allows to achieve recognition accuracy equal to the algorithm with the best parameters set, and not only for Kullback-Leibler information discrimination, but also for other popular distances (Euclidean metric, Kullback information divergence).

Added: Jun 9, 2012
Article
Евсютин О. О. Компьютерная оптика. 2014. Т. 38. № 2. С. 314-321.

This paper is aimed at receiving orthogonal bases families from the evolving states of cellular automata. I suggest a comparison technique of the appropriate orthogonal transformations in respect of noises, shown as a result of information losses on the restored data elements.

Added: Sep 5, 2019
Article
Савченко А. В. Компьютерная оптика. 2017. Т. 41. № 3. С. 422-430.

In this paper we focus on the image recognition problem in the case of small sample size based on the nearest neighbor rule and matching of high-dimensional feature vectors extracted with the deep convolutional neural network. We propose the novel recognition algorithm based on the maximum likelihood method for the joint density of dissimilarities between an observed image and available instances in the training set. This likelihood is estimated using the known asymptotically normally distribution of the Jensen-Shannon divergence between image features, if the latter can be treated as the probability density estimates. This asymptotic behavior is in agreement with the well-known experimental estimates of distributions of dissimilarity distances between high-dimensional vectors. The experimental study in unconstrained face recognition for the LFW (Labeled Faces in the Wild) and YTF (YouTube Faces) datasets demonstrated, that the proposed approach makes it possible to increase the recognition accuracy at 1-5% when compared with conventional classifiers.

Added: Jul 8, 2017
Article
Евсютин О. О., Кокурина А. С., Мещеряков Р. В. Компьютерная оптика. 2019. Т. 43. № 1. С. 137-154.

Transmission, processing and storage of information in the infrastructure of the Internet of Things are related to the necessity for solving a number of problems in information security. The main difficulty lies in the fact that the infrastructure of the Internet of Things is not homogeneous and contains many different devices, including those with limited computing resources. One of the approaches to solving these problems is to embed additional information into the transmitted and stored digital objects. In this paper we present a review of methods of embedding information in digital data to provide security in the Internet of Things, including methods of steganographic embedding of information and methods for embedding digital watermarks. We reviewed methods of embedding information into digital images, as well as wireless sensor network data, proposed for use in the Internet of Things. In this paper we defined the advantages and disadvantages of individual methods and groups of methods, also we analyzed their applicability for data protection in the Internet of Things. Relevant trends in this field of research have been identified.

Added: Sep 3, 2019
Article
Курочкин С. В. Компьютерная оптика. 2019. Т. 43. № 4. С. 611-617.

A method of topological data analysis is proposed that allows one to find out the homotopy type of the object under study. Unlike mature and widely used methods based on persistent homologies, our method is based on computing differential invariants of some map associated with an approximating map. Differential topology tools and the analogy with the main result in Morse theory are used. The approximating map can be constructed in the usual way using a neural network or otherwise. The method allows one to identify the homotopy type of an object in the plane because the number of circles in the homotopy equivalent object representation as a wedge is expressed through the degree of some map associated with the approximating map. The performance of the algorithm is illustrated by examples from the MNIST database and transforms thereof. Generalizations and open questions relating to a higher-dimension case are discussed.

Added: Oct 1, 2019
Article
Фурсов В. А., Козин Н. Е. Компьютерная оптика. 2008. Т. 32. № 4. С. 400-402.
Added: Feb 7, 2010
Article
Шахуро В. И., Конушин А. С. Компьютерная оптика. 2016. Т. 40. № 2. С. 294-300.

A new public dataset of traffic sign images is presented. The dataset is intended for training and testing the algorithms of traffic sign recognition. We describe the dataset structure and guidelines for working with the dataset, comparing it with the previously published traffic sign datasets. The evaluation of modern detection and classification algorithms conducted using the proposed dataset has shown that existing methods of recognition of a wide class of traffic signs do not achieve the accuracy and completeness required for a number of applications.

Added: Jul 8, 2016
Article
Шахуро В. И., Конушин А. С. Компьютерная оптика. 2018. Т. 42. № 1. С. 105-112.

In this work, we research the applicability of generative adversarial neural networks for generating training samples for a traffic sign classification task. We consider generative neural networks trained using the Wasserstein metric. As a baseline method for comparison, we take image generation based on traffic sign icons. Experimental evaluation of the classifiers based on convolutional neural networks is conducted on real data, two types of synthetic data, and a combination of real and synthetic data. The experiments show that modern generative neural networks are capable of generating realistic training samples for traffic sign classification that outperform methods for generating images with icons, but are still slightly worse than real images for classifier training.

Added: Oct 31, 2019
Article
Фурсов В. А., Козин Н. Е. Компьютерная оптика. 2008. Т. 32. № 3. С. 307-310.
Added: Feb 7, 2010
Article
Савченко А. В. Компьютерная оптика. 2018. Т. 42. № 1. С. 149-158.

In this paper we study the image recognition tasks, in which images are described by high dimensional feature vectors extracted with deep convolutional neural networks and principal component analysis. In particular, we focus on the problem of high computational complexity of statistical approach with non-parametric estimates of probability density implemented by the probabilistic neural network. We propose the novel statistical classification method based on the density estimators with the orthogonal expansions using trigonometric series. It is shown that this approach makes it possible to overcome the drawbacks of the probabilistic neural network caused by the memory-based approach of instance-based learning. Our experimental study with Caltech-101 and CASIA WebFaces demonstrates that the proposed approach reduces error rate at 1-5%, and increases computational speed in 1.5-6 times when compared to the original probabilistic neural network for small samples of reference images.

Added: Apr 11, 2018
1 2