Tree-like data structures for effective recognition of 2-D solids

An approach to extracting and recognizing the patterns given by the solids in two-level images is developed. The applied data processing and analyzing procedures are based on using the tree-structured representations of both the whole image and the individual patterns. These representations provide a significant reduction of calculations at the stages of pattern segmentation and recognition. The estimations of the fidelity and computational complexity performances have been obtained for the recognition procedure based on source coding and sequential decoding schemes.