Abstract: Aiming at the problem that it is hard to synthesize handwritten Chinese character, this paper proposes a new method of handwritten Chinese character synthesis based on the structure knowledge of off-line Chinese character. It uses improved direction distance length algorithm to extract stokes of character and establishes stoke bounding boxes, synthesizes handwritten Chinese character based on the knowledge of stoke bounding boxes. Experimental result shows that the method is effective for generating some Chinese characters, these characters are easy to be recognized by human and their resistances to OCR(Optical Character Recognition) are strong.

Key words: stoke extraction, direction run length algorithm, structure knowledge of Chinese character, stoke relationship matrix, Chinese character synthesis

摘要： 针对手写体汉字合成困难的问题，从离线手写体汉字样本的结构知识出发，提出一种合成手写体汉字的方法。该方法采用改进的方向游程算法提取汉字的笔画，并建立笔画框，从笔画框的结构知识出发构建起字体信息来合成汉字。实验结果证明，该方法可以较好地合成汉字，且容易识别，对光学字符识别有较好的抵抗能力。

关键词: 笔画提取, 方向游程算法, 汉字结构知识, 笔画关系矩阵, 汉字合成

CLC Number: 

• TP391