RECOGNITION OF HANDPRINTED CHINESE CHARACTERS BY CONSTRAINED GRAPH MATCHING

A model-based handwritten Chinese character recognition (HCCR) system is proposed. The characters are represented by attributed relational g raphs (ARG) using strokes as ARG vertices. A number of vector relation al attributes are also used in the representation to improve the perfo rmance of the translation and scale invariant and rotation sensitive r ecognition system. Since the ETL-8 database is very noisy and broken s trokes are commonly encountered, a suitable homomorphic energy functio n is proposed that allows the segments of a broken stroke of a test ch aracter to be matched to the corresponding model stroke. The homomorph ic ARG matching energy is minimised using the self-organising Hopfield neural networks [1] [Suganthan, P.N., Teoh, E.K., Mital, D.P., A self -organising Hopfield network for attributed relational graph matching, Image and Vision Computing, 13(1) (1995) 61-73]. An effective formula tion is introduced to determine the matching score. The formulation do es not penalise the matching scores of test characters with broken str okes. Experiments were performed with 100 classes of characters in the ETL-8 database and 98.9% recognition accuracy has been achieved. (C) 1998 Elsevier Science B.V.