摘要

A structural algorithm for recognition of objects in skeletonized drawings and graphic images is presented. The algorithm proposed uses a grammar in order to assign the set of structural descriptions of objects and detects the elements of these objects in the process of their recognition under the control of the grammar. The algorithm was realized in the system for automatic input, processing and semantic interpretation of drawings and graphic images. In order to recognize new objects, one has only to specify their grammars without rewriting the recognition program. The results of experiments dealing with recognition of schematic circuit diagrams, maps, and images of handwritten figures confirm the high processing speed, efficiency, and flexibility of the algorithm.