Agarkov A. Structural image description and recognition through the use of graphs

Grayscale bitmaps and the means of presentation, description, analysis and processing are scrutinized the In dissertation. The aim of the thesis is to improve the efficiency of algorithms for recognition of images by combining the structure and shape. This thesis had been applied graph theory, elements of set theory, digital imaging and digital transformation of two-dimensional information for the transformation of the input images, and fast convolution calculation, statistical treatment of experimental data. It was developed a means of structural description and recognition of grayscale images which consider both of the structure and shape of objects. This one making it possible to increase the information content of image descriptions in comparison with analogues. On the basis of these tools developed an experimental technique to detect and recognize objects, which is in contrast to similar equally effective in recognition of textured and non-textured objects, allows the detection of identical objects without a priori defined templates. The results of the dissertation research used in the development of software and in the educational process, designed to create a system of visual data mining.