A hierarchical pattern description in the syntactic approach to pattern recognition

Abstract

A variable selectivity description method for syntactic pattern recognition is proposed in this paper. The method is based on the possibility of a suitable presentation of patterns in a recognition problem. Hierarchical description systems are used to specify the pattern classes. In these systems the generation of words proceeds successively on separate levels by the parallel mode of rewriting. The recognition procedure utilizes the considered hierarchical pattern description. In consequence of that the processing of some patterns can be speeded. The hierarchical description systems was used in the numerical character recognition experiment. In this paper a short mention of that is given to illustrate the proposed method.