摘要
针对汉字识别的超多类问题,将贝叶斯网络分类器引入小样本字符集脱机手写体汉字识别中.对手写大写数字汉字的小样本字符集构造识别系统,同时与传统的欧氏距离方法进行比较,实验表明该算法将识别率提高到92.4%,在小样本字符集脱机手写体识别中具有较强的实用性和良好的扩展性.
With the super-multi-class issue in Chinese character recognition, Bayesian network classifier is introduced into small-set offline handwritten Chinese character recognition, for which a recognition system is constructed and it is compared with Euclidean distance classifier. The experiments indicate that it can increase the recognition rate to 92.4% , and therefore it has more practicability and scalability.
出处
《计算机辅助工程》
2006年第3期72-74,89,共4页
Computer Aided Engineering
基金
教育部人文社会科学研究规划项目(2005-241)
广东省科技攻关项目(2005B10101010)
关键词
贝叶斯网络
分类器
脱机手写体汉字
智能识别
欧氏距离
Bayesian network
classifier
offline handwritten Chinese character
intelligent recognition
Euclidean distance