Vol 27, No 4 (2017)

This paper is devolved to descriptive image analysis, an important, if not a leading, direction in the modern mathematical theory of image analysis. Descriptive image analysis is a logically organized set of descriptive methods and models meant for analyzing and estimating the information represented in the form of images, as well as for automating the extraction (from images) of knowledge and data needed for intelligent decision making about the real-world scenes reflected and represented by images under analysis. The basic idea of descriptive image analysis consists in reducing all processes of analysis (processing, recognition, and understanding) of images to (1) construction of models (representations and formalized descriptions) of images; (2) definition of transformations over image models; (3) construction of models (representations and formalized descriptions) of transformations over models and representations of images; and (4) construction of models (representations and formalized descriptions) of schemes of transformations over models and representations of images that provide the solution to image analysis problems. The main fundamental sources that predetermined the origination and development of descriptive image analysis, or had a significant influence thereon, are considered. In addition, a brief description of the current state of descriptive image analysis that reflects the main results of the descriptive approach to analysis and understanding of images is presented. The opportunities and limitations of algebraic approaches to image analysis are discussed. During recent years, it was accepted that algebraic techniques, particularly, different kinds of image algebras, are the most promising direction of construction of the mathematical theory of image analysis and of the development of a universal algebraic language for representing image analysis transforms, as well as image representations and models. The main goal of the algebraic approaches is designing a unified scheme for representation of objects under recognition and its transforms in the form of certain algebraic structures. This makes it possible to develop the corresponding regular structures ready for analysis by algebraic, geometrical, and topological techniques. The development of this line of image analysis and pattern recognition is of crucial importance for automatic image mining and application problems solving, in particular, for diversification of the classes and types of solvable problems, as well as for significant improvement of the efficiency and quality of solutions. The main subgoals of the paper are (1) to set forth the-state-of-the-art of the mathematical theory of image analysis; (2) to consider the algebraic approaches and techniques suitable for image analysis; and (3) to present a methodology, as well as mathematical and computational techniques, for automation of image mining on the basis of the descriptive approach to image analysis (DAIA). The main trends and problems in the promising basic researches focused on the development of a descriptive theory of image analysis are described.

In this paper we consider the problem of removing achromatic reflections from a picture of a scene taken through a semi-transparent medium, assuming that the reflection pattern is due to a light source or another object located in front of the object of interest. While other works assume the availability of multiple observations, we consider the more challenging problem of having as data a single color image. We suppose a data model where the virtual reflected image combines additively with the real transmitted image of the object, through unknown coefficients. This highly underdetermined problem is handled by means of a blind estimation technique that exploits the strict dependence of the gradients of the three color channels of the ideal image, and their independence from the gradient of the grayscale reflected image. The model parameters are estimated through independent component analysis, and then the component images are estimated through a regularization technique. The whole algorithm is very fast, and its performance is quantitatively evaluated on numerically generated images, and qualitatively tested on real images.

Image enhancement can be formulated as an optimization problem where one parameterized transformation function is used for enhancement purpose. The proper enhancement significantly depends on two factors- fine tuning of the parameters of the corresponding parameterized transformation function and other one is the selection of a proper objective function. In this study a parameterized variant of histogram equalization (HE) has been used for enhancement purpose and to tune the parameters of that variant a modified cuckoo search (CS) with new global and local search strategies is employed. This paper also concentrates on the selection of a proper objective function to preserve the original brightness of the image. A new objective function has been developed by combining fractal dimension (FD) and quality index based on local variance (QILV). Visual analysis and experimental results prove that modified CS with search strategies outperforms the traditional and some other existing modified CS algorithms. Considering the image’s brightness preserving capability, the proposed objective function significantly outperforms other existing objective functions.

Content-based image retrieval system based on global visual content features normally return the retrieval results according to the similarity between features extracted from the sample query image and candidate images. However, global features usually cannot capture different characteristics of different parts in the image. Therefore, the representation of local image properties is one of the most active research issues in content-based image retrieval. The method based on salient point detection is one of the typical and effective approaches. This paper proposes three improved salient point detectors based on wavelet transform, which are calculated in the three different orientations’ and scales’ subbands and weighted equally. In contrast to the former method based on salient point detection, the improved salient point detectors aim to extract the visual information in the image more effectively. We have tested the proposed schemes and compared four salient point detectors using a wide range of image samples from the Corel Image Library, and experimental results show that the improved salient point detectors have produced promising results.

Double threshold method of traditional Canny operator detects the edge rely on the information of gradient magnitude, which has a lower edge connectivity and incomplete image information. Aiming at this problem, we proposed an edge detection algorithm based edge connection—the Hough Transform based Canny (HT-Canny) edge detection algorithm. HT-Canny algorithm guided by high threshold image, which obtains edge direction through calculating edge endpoint gradient and connects the edge by using the Hough Transform instead of traditional double threshold method. It avoids the limitation of traditional Canny algorithm, which must set the double threshold manually and protect the low intensity edge especially. The experimental results show that HT-Canny algorithm has stronger edge connectivity and can distinguish edge points and non-edge points effectively, which not only retain the advantages of the traditional Canny algorithm but also make the detection result more complete and comprehensive.

In this work we discuss methods for image ringing detection and suppression that are based on the sparse representations approach and suggest a new ringing suppression method. The ringing detection algorithm is based on construction of the synthetic dictionary that is used to represent ringing effect as a sum of blurred edge and pure ringing component. This decomposition enables us to estimate image ringing level. We analyze two ringing suppression methods. First method is based on learning joint dictionaries and shows good performance for the whole image on average. However for high ringing levels the performance of this method decreases due to the influence of the ringing artefact on the sparse representation parameters. The second method is based on separate learning of natural images dictionary and pure ringing dictionary and it does not suffer from this problem. In this article we present a new ringing suppression method that is based on the method using separate dictionaries. The method works best in the areas of edges and for higher levels of ringing effect.

This paper provides a thorough evaluation of a set of six important Arabic OCR systems available in the market; namely: Abbyy FineReader, Leadtools, Readiris, Sakhr, Tesseract and NovoVerus. We test the OCR systems using a randomly selected images from the well known Arabic Printed Text Image database (250 images from the APTI database) and using a set of 8 images from an Arabic book. The APTI database contains 45.313.600 of both decomposable and non-decomposable word images. In the evaluation, we conduct two tests. The first test is based on usual metrics used in the literature. In the second test, we provide a novel measure for Arabic language, which can be used for other non-Latin languages.

The paper is devoted to the extension of the intelligent systems complex for express-diagnostics consisting of express-diagnostics systems of organizational stress (DIOS), depression (DIAPROD), and deviant behavior (DIDEV) by the addition of a system of express-diagnostics designed for diagnosis of anxiety (DIAA). Application of the extended complex of intelligent systems of express-diagnostics of organizational stress, depression, deviant behavior, and anxiety (ISED OSDDA) will make it possible to promptly make diagnostic, preventative medical, and organizational-management decisions on the aforementioned disorders on the basis of the biopsychosocial approach; to justify these decisions with the use of graphic (including cognitive) tools; and to make complex decisions on the subject under investigation.

The study of the consequences of traumatic brain injury represents an important problem in modern medicine. Brain trauma leads to the manifestation of neurological and psychiatric disturbances in most patients; severe traumatic brain injuries are accompanied by paroxysms and neurologic impairment, which in the long term lead to post-traumatic epilepsy. The mechanisms of post-traumatic epilepsy, in view of cellular and molecular plasticity, include rearrangement of neuronal networks. These processes induce spontaneous and synchronous discharges of numerous neurons, clinically manifested in the form of paroxysms. The posttraumatic period before the first unprovoked seizures is known as the latent phase of post-traumatic epileptogenesis, which can last for years in humans. Unfortunately, approaches to its detection and to the prediction of post-traumatic epilepsy have not been sufficiently developed yet; though they are of exceptional importance for clinical practice. Moreover, it is still poorly understood how the damage of the brain tissue leads to the brain structural rearrangement connected with epileptogenesis. In other words, the question is why and when the epileptogenesis occurs. The current paper considers the first results of the approach that we have suggested to quantitative assessment of the epileptogenesis development on the basis of EEG analysis as an alternative to the regular activity of the brain. This approach is based on the analysis of highly nonstationary signals that contain quasiperiodic fragments. In particular, the pipeline of procedures for the analysis includes detection of local quasiperiodic (rhythmic) behavior with further assessment of its dynamic characteristics and segmentation of fragments with “rhythmic” behavior. The first results of this approach application are connected with the analysis of experimental data obtained in the modeling of traumatic brain injury in rats. The characteristics of the EEG rhythm and their dynamics are analyzed in the background (before traumatic brain injury) and one, two, three and six days after the injury on the basis of ten-hour EEG recordings. The results demonstrate a significant difference in the degree of periodicity of the characteristics (total duration of the corresponding EEG fragments) before and after traumatic brain injury (a week after the trauma some restoration of the periodicity is observed). At the end of the paper we offer some interpretation of these results and express a hope that such studies could provide new ways for development and assessment of methods of the paroxysmal activity in the post-traumatic period.

The property of rotation, scaling and translation invariant has a great important in 3D image classification and recognition. Tchebichef moments as a classical orthogonal moment have been widely used in image analysis and recognition. Since Tchebichef moments are represented in Cartesian coordinate, the rotation invariance can’t easy to realize. In this paper, we propose a new set of 3D rotation scaling and translation invariance of radial Tchebichef moments. We also present a theoretical mathematics to derive them. Hence, this paper we present a new 3D radial Tchebichef moments using a spherical representation of volumetric image by a one-dimensional orthogonal discrete Tchebichef polynomials and a spherical function. They have better image reconstruction performance, lower information redundancy and higher noise robustness than the existing radial orthogonal moments. At last, a mathematical framework for obtaining the rotation, scaling and translation invariants of these two types of Tchebichef moments is provided. Theoretical and experimental results show the superiority of the proposed methods in terms of image reconstruction capability and invariant recognition accuracy under both noisy and noise-free conditions. The result of experiments prove that the Tchebichef moments have done better than the Krawtchouk moments with and without noise. Simultaneously, the reconstructed 3D image converges quickly to the original image using 3D radial Tchebichef moments and the test images are clearly recognized from a set of images that are available in a PSB database.

This paper studies the application of N-grams for correcting the results of pattern recognition of words in documents based on the example of recognition of passport fields of a citizen of the Russian Federation. Three algorithms for correcting recognition results are given for trigrams. One of them is based on the use of trigram probabilities in combination with evaluation of recognition. The other algorithms are based on the definition of marginal distributions and computations by means of graphical probability models. The results of experiments on the application of the algorithms and comparison of the characteristics of the algorithms are presented.

Locating the center of the eyes plays a significant role in many computer vision applications and research, such as face alignment, face recognition, human-computer interaction, control devices for disabled people, user attention and gaze estimation. The disturbances such as occlusions by eyelashes or eyelids, uneven spots and spectacle frames of glasses affect the accuracy and stability of eye center location. This paper presents a hybrid eye center locating methodology for infrared eye images. The pupil edge points are extracted by Starburst algorithm, and when we get the position and the gradient of the edge points, the approximate pupil boundary is determined by a convex region voting methods. After that, the boundary edge points are iteratively optimized by fitting an ellipses modeling constraint. Finally, the pupil is located correctly. Experiment shows that this algorithm has performance advantages compared with some state of the art approaches in pupil localization accuracy, iteration times and their performance. This algorithm combining convex area voting and model constraint has strong robustness, high accuracy and speed in real environments with occlusions and distortion pupil.