STRUCTURAL ATTRIBUTE FEATURE CODE REPRESENTATION AND RECOGNITION OF MULTIFONT PRINTED CHINESE CHARACTERS
Z. Lin, Ping Liu · International Journal of Pattern Recognition and Artificial Intelligence · 2001
In this paper, a new structural representation and fuzzy matching scheme are proposed for multifont printed Chinese character recognition. A Chinese character is decomposed into eight stroke types. A complete structural attribute feature codes among different types of strokes are defined and extracted, which consist of weak and strong primary codes and secondary codes. Weak and strong primary feature codes depict the global and local spatial relationships among different types of strokes respectively, and they are used for a detailed match. A fuzzy matching scheme is used for detailed match between an input character and candidate characters. An experiment on 3755 Chinese characters used daily in multifonts and multisizes shows that our method is robust and can achieve high recognition accuracy.