Da Lu

An Approach to Handwritten Chinese Character Preclassification Based on Fuzzy Logic and Similarity Measure

This paper presents a fuzz logic approach and similarity measure to efficiently perform handwritten Chinese character preclassification. After extracting stroke density features and using fuzzy rules to create some preclassification groups, we transform the characters in each group into fuzzy prototypes based on a nonlinear weighted similarity function by fuzzy logic approach, then we use matching algorithm and hierarchy classification of fuzzy prototypes of similarity measurement for character preclassification. Experiment shows that our method is effective, and the characters of the testing set can be distributed into correct preclassification classes and a preclassification rate of 98.17% is achieved

Study on preclassification for handwritten Chinese character based on neural net and fuzzy matching algorithm

To settle the recognition task of handwritten Chinese characters, the authors put forward a method for handwritten Chinese character preclassification before character recognition. In this method, Neocognitron was used in extracting stroke features, then uses the Supervised Extended ART (SEART) to create some preclassification groups, and uses matching algorithm of fuzzy prototypes of similarity measurement for character preclassification. The experiment shows this method is effective when used for handwritten Chinese character classification and characters of the testing set can be distributed into correct preclassification classes at a rate of 98.22%.