A Chinese character recognition interface for mobile communication devices using fuzzy logic and unit extraction

A novel method for recognition of on-line handwritten Chinese characters is presented with the emphasis to a systematic representation in fuzzy logic for characters. The information contained in a character is examined by a set of well-selected features. A matrix structure is defined as an abstract fuzzy representation for the characters determined by inference of fuzzy rule base. Chinese characters can be coded in a small set of basic units. Hence, a unit extraction module is designed to search the possible units within characters. The input character is recognized by the unit code with the largest fuzzy similarity degree. The training capability is performed with the qualitative fuzzy c-means clustering algorithm. An experimental system has achieved a recognition rate of 95% under a test set of 550 characters and 2 cases for each character written by 6 users. The experimental system has real-time recognition response of average 0.5 sec/character. This approach has two significant advantages over other methods: (1) less training time; and (2) less storage required to store character models, that are two crucial factors in design of personal digital assistant (PDA) devices for mobile communication. The storage using our method requires O(i/sup 2/) memory space, where i is the average number of strokes in a unit. In contrast, the method storing information of entire character requires O(I/sup 2/) memory space, where I is the average number of strokes in a character.