Babilunga O. Iterative methods of the hierarchic contour segmentation images at the hyperbolic wavelet-transformation space

Dissertation is devoted to the increasing speed of intelligence processing and pattern recognition systems by development of iterative methods of the hierarchical contour segmentation images at the hyperbolic wavelet transformation space. The model of description hierarchical structure objects with the semantic means is offered. The iterative search method on the basis of hyperbolic wavelet-transformation (HWT) and the iterative edge detection methods of gray scale images are developed. Speed increase of the edge describing gray scale images by the iterative methods was reached. The morphological edge processing method of gray scale images by the iterative methods is considered. It has allowed increasing speed of the edge describing and uniting edge detection, skeletonization and edge tracing. The technique of the classification of the hierarchical structure objects in HWT space is developed. It has allowed to receive the description of complex structure object at the different hierarchical levels andto increase speed of intelligence processing and pattern recognition systems at 1,3 – 5 times.