一种活动模板子结构引导的联机手写汉字识别方法

结构匹配是一种有效的联机手写汉写识别方法,为了减少匹配运算,人们一直在寻 求利用部分匹配的结果来引导整体匹配的方法.在特征匹配与结构匹配综合的基础上,从 3.755个一级国标汉字中提取出45个子结构,利用它们来引导结构匹配.由于这些子结构总 出现在字首或字尾,因而对它们的检测比较容易.同时,通过建立子结构活动模板及设计子结 构动态抽取算法,使得子结构匹配的准确度得到很大提高.实验结构表明,该方法使结构匹配 的运算量减少约50％,并对类似的物体识别问题有一定的启发意义.

On-Line Handwritten Chinese Character Recognition Guided by Components with Dynamic Templates

Structural matching is a main approach for on line Chinese character recognition.In order to reduce its great computational complexity and improve its performance,people have been seeking for a way to guide the whole matching by the result of partial matching.In this paper,the authors proposed 45 basic components from 3,755 categories of the daily used Chinese characters to guide the stroke segment matching.Because they always locate at either the beginning or the end of the stroke segment string,these components are easy to extract and separate from other parts of the character.Besides,the reference templates of these components are dynamically extracted from the reference segment string and dependent on the current matched character so that a more accurate matching is carried out.Experiments show that the segment matching computation has reduced almost 50%.The approach is also enlightening for other similar object matching problem.