A Partial Area Matching Method Based on Support Vector Machine for Distinguishing Similar On-Line Handwritten Chinese Characters

In this paper, a new partial area matching method based on support vector machine (SVM) for distinguishing similar on-line handwritten Chinese characters is presented. With this new method, the similar characters are divided into several classes according to the difference among their structure features. After the candidate character set is obtained by the former recognizing steps, the SVM is used to pick up the most accurately matched character from the set. Experiments showed that the similar Chinese characters can be distinguished effectively and the recognition rate for the first rank Chinese characters can be improved with this method.