Multicomponent Model of Objects Attributive Signatures on Color Pictures

A technique for multicomponent model synthesis aimed at description of color, geometric and structural attributes of objects images on pictures is suggested. In terms of the model an object is represented as a hierarchical set of embedded contours, wherein the upper level of the object topological representation is the description of the external boundary of that object image and the lower levels of the topology detailing correspond to the internal structure elements. For the contours detection color information of the picture statistically homogeneous regions is used. To select the external contour points an algorithm for forming the contour circular sweep with the uniform angle step as well as the methods for binding contour points, which are based on the criterion of distance between the neighboring points of the contour have been developed. The presented results testify the possibility of using the suggested technique in the procedures of recognition of graphical objects.