Oliynyk O. . Models and methods for segmentation of irregular image bounds on the basis of adaptive edge detectors

The thesis is devoted to development of models and methods for solving of problems of segmentation of par-tially contrast irregular image bounds on the basis of adaptive masks and edge detectors. A system of models is proposed that permits to take account of brightness, con-trast, topological and geometrical image parameters. On these grounds the general image bound segmentation problem is reduced to the main problem of bound nodes segmentation and construction of image bound compo-nents, and to a series of auxiliary problems. For solving of the main problem a system of adaptive masks and edge detectors is proposed and a series of criteria that provide adequacy, completeness and stability of segmentation; for the masks, edge detectors and criteria a system of tem-plates and the methods for adjusting their parameters are developed. A scheme for solving the general problem of irregular image boundary segmentation is proposed for the systems of technical vision.