摘要
本文针对汉字楷体手书识别问题,从整体计算的角度,提出了一种汉字形体特征提取的解决方法并付诸于具体系统的实现。由于强调了汉字整体信息和相对关系的结构特征,比较成功地把握了区分手书汉字的关键因素,汉字识别率达80%以上,效果比较理想。
Aiming at the recognition problem for Regular Script Chinese Character (RSCC), we have advanced a new method for abstracting the features of the RSCC from the view of entire evolution and have constructed a computer recognition system. Becausce we emphasized the entire information, relative relationship between components in our method, we have obtained & good effect and the rate of correct recognition for RSCC comed to 80%.
出处
《电子学报》
EI
CAS
CSCD
北大核心
1993年第2期84-85,共2页
Acta Electronica Sinica
基金
北京大学视觉
听觉开放实验赞助项目
关键词
楷书识别
特征提取
计算
汉字识别
Feature abstracting, Computation in entirety, Handwriting recognition