Zahorodnia D. Informational Technology for Structural and Statistic Identification of Hierarchical Objects from Characteristic Points of their Contours in Automated Video Surveillance Systems

Dissertation is devoted to the development of contour segmentation methods based on the Kanny's algorithm and identification of contour characteristic points for improving performance of automated video surveillance systems within limited computing resources. The method and basic information technology of the contour segmentation on the basis of Kanny's method was developed in the dissertation, a distinctive feature of which is usage as underlining wavelet transformation functions which allowed to obtain a sequence of the hierarchical object contour items with the detailing that can be regulated; the method and basic information technology of the contour inflection points selection with the objects' regulated detailing were developed, in the basis of which resides the wavelet analysis of the curvature function which allowed to diminish the amount of the contour inflection points in accordance to the goal of the processing. The method and basic information technology of the structure and statistic object identification on the basis of identification vectors with hierarchical sequence implementation calculated on the basis of the object's contour inflection points with a regulated detailing was improved, it allowed to increase working speed of the whole automated video surveillance system with the high authenticity. On the basis of the developed methods the information technology of the structural and statistical identification of hierarchical objects by the contour characteristic points was created, it was implemented in automated computer video surveillance systems.