D. M. Murashov

A two-level method for segmenting cytological images based on active contour model

The problem of automatic extraction of cell nuclei in cytological images for automated diagnostics of hematological diseases is considered. A combined two-level segmentation method based on an active contour model is presented. The method employs different strategies to form the active contour model at coarse and fine approximation steps. The capture range of the active contour is extended by employing a wave propagation model. Unsupervised initialization of the active contour involves binarization of the a component of the input image in the CIE Lab color space. The efficiency of the developed method is demonstrated by a computational experiment.

Localization of differences between multimodal images on the basis of an information-theoretical measure

The problem of the localization of image artifacts obtained in different spectral ranges on the basis of an information-theoretical difference measure is considered. Use of the conditional entropy value, calculated on a pair of images, is proposed for the measure of difference. The applicability conditions of this measure are specified and its testing is carried out. An example of the application is presented, using the difference measure to localize the repainting and retouching of areas in the images of fine-art paintings in the visible and ultraviolet bands.

An open general-purposes research system for automating the development and application of information technologies in the area of image processing, analysis, and evaluation

Algorithmic/software system (ASS) referred to as “Black Square” (BS) is described. It is intended for scientific research and education in the area of image processing, analysis, recognition, and understanding. This ASS is a software environment for designing and application of algorithms for image processing, analysis, recognition and understanding with reference and information-retrieval functions. The ASS operates by means of a unified integrated software environment constructed on a common informational and methodical basis with a common navigation shell. The minimal specification necessary for launching the BS is the following: Pentium III processor 700 MHz; 128 Mb RAM; at least 100 Mb of free space on the hard disc; and Windows 98/ME/2000/XP or Linux RedHat 9. The ASS is intended for solving the following problems: (1) education and knowledge accumulation in the area of image processing, analysis, and recognition; (2) development, testing, and storage of algorithms and information technologies for solving problems of image processing, analysis, and recognition; (3) design of problem-oriented ASSs for image analysis and recognition; (4) automation of design, investigation, and application of algorithms for image processing, analysis, recognition, and understanding; (5) automation of training for methods for image processing and analysis; and (6) well-structured and user-friendly storage of various information, including information on methods and algorithms for image processing, analysis, recognition, and understanding, as well as images themselves and information about images. The main areas of application of the BS are the following: automation of scientific research and investigations in the area of image recognition and analysis; development of information technologies; solution of applied problems in the field of medicine (diagnoses of various diseases, in particular, cancer; investigation of the influence of laser and magnetic radiation on human beings; investigation of the influence of various factors on the temperature field of a human being); aerospace survey (analysis of data of remote sensing); development of tracking and detection systems; industrial and power engineering (technical diagnostics and nondestructive monitoring); as well as ecological monitoring and emergency prediction and monitoring. As an object of intellectual property, the work is protected by two patents and two inventor’s certificates.

Feature description of informative fragments in the problem of computerized attribution of paintings

Problems of composing a feature description and developing a comparison procedure for the images of paintings in attribution are considered. A feature description of a facture of paintings based on the characteristics of a grayscale image relief and elements of the structure tensor is proposed. In contrast to known techniques, the feature description is formed only by informative fragments of images and does not require preliminary segmentation of individual brushstrokes. Parameters of feature extraction procedures are chosen. Measures of dissimilarity between images of paintings are proposed. Computational experiments are carried out. The feature description proposed is a quantitative characteristic of the artistic style of an author. The procedure developed for comparing images can be applied together with other types of investigation of paintings to make an attributional conclusion.

Comparative study of algorithms for matching sets of reference points

Experimental investigation of the algorithms for matching the sets of reference points in the problem on registration of images of fine art paintings is presented in this paper. The experiments are carried out using synthetic and real data sets. The decision about the possibility of applying the examined algorithms to image registration is made. The most suitable algorithm is chosen, and the development of a procedure for eliminating false correspondences is stated to be vital.

Compensation for local defects in an image created using a triple-color photo technique

Problems connected with the creation of elements of technology for automated detection and the correction of image local defects obtained in the triple-color photo technique are considered. Automated procedures of detection and correction are developed. Procedures are implemented in software-tool and used in works on the reconstruction of S. M. Prokudin-Gorskii’s collection of photos taken in the early 20th century.

Application of texture features for comparing the facture of paintings

The paper continues investigations on the development of a computer-aided method for the analysis of images of the facture of pictorial artworks. The feature description of the facture of paintings is extended. An earlier developed method of comparing a facture by features extracted from informative fragments of digitized images of portraits is applied. The features describe the direction of a brushstroke, which characterizes the artistic manner specific for specific details of a painting. A numerical experiment has shown that the value of the quantitative similarity index of paintings of the same artist (F.S. Rokotov) is higher than the value of the quantitative similarity index of the works of different artists. A portrait attributed to Rokotov is compared with his authentic works, as well as with portraits painted by different artists of the XVII−XIX centuries. The results of the computer analysis of the facture of paintings do not contradict the results of traditional technico-technological investigations.

Experimental study of the evolution of scale-space cytological images

This paper covers an experimental study of the evolution of the features of families of blurred images of cell nuclei. The purpose of this study was to identify new diagnostic criteria for solving the problem of automated differential diagnostics using the images of cytological preparations. New diagnostic criteria in the aspect of the dependences of the number of brightness extrema on the scale are presented; the description of nuclei images in the distribution of the nucleus area versus brightness local minimum amount at various scale values is obtained.