PATTERN-RECOGNITION BY HOMOMORPHIC GRAPH MATCHING USING HOPFIELD NEURAL NETWORKS

The application of the Hopfield neural network as a constraint satisfa ction network for pattern recognition is investigated in this paper. S uitable energy and compatibility functions are introduced for pattern recognition by homomorphic attributed relational graph (ARG) matching. Although many computer vision problems have been traditionally formul ated as combinatorial optimization problems, most of them can be reduc ed to that of finding the nearest local minimum of an objective functi on. In this paper, a novel network initialization strategy is applied to achieve the desired complexity reduction. Further, a method to veri fy and localize the hypotheses generated by the Hopfield network is al so presented using an efficient pose clustering algorithm. The perform ance of the connectionist approach to pattern recognition by homomorph ic relational graph matching is demonstrated using a number of line pa tterns, silhouette images and circle patterns.