脱机手写满文笔画基元的提取和识别  

Separation and Recognition Method for Off-line Handwritten Manchu Character Strokes

作　　者：张广渊[1] 李晶皎[2] 王爱侠[2] 

机构地区：[1]清华大学汽车工程系,北京100084 [2]东北大学信息科学与工程学院,沈阳110004

出　　处：《计算机工程》2007年第22期200-202,共3页Computer Engineering

基　　金：国家自然科学基金资助项目(69683004);辽宁省自然科学基金资助项目(2001113)

摘　　要：笔画的正确提取对文字识别结果至关重要。该文提出并实现了一种全新的满文笔画提取方法。基于满文文字的结构和特征,首先对处理目标进行预处理,确定笔画主干,创建笔画生长法对满文文字笔画实现自动提取。对提取得到的笔画进行分类,然后对照笔画特征库进行识别。试验表明,该方法对规整手写的满文文字的笔画提取和识别有令人满意的效果,为进一步脱机手写满文识别研究奠定了坚实的基础。Correct strokes separation has great influence on the recognition result. This paper elaborates on the method separating and recognizing off-line handwritten Manchu character strokes. Based on the structure and characteristic of Manchu, the character is pre-processed. Then the mid-axis of Manchu character is defined and stroke growth algorithm is created to separate the strokes. The separated stroke is recognized by comparing its characteristic with the standard stroke characteristic. Simulation results show that the method is effective in separating and recognizing restricted off-line handwritten Manchu characters.

关 键 词：满文 脱机手写字符 笔画 模式识别 

分 类 号：TP391.43[自动化与计算机技术—计算机应用技术]