摘要
多分类器集成是解决手写体汉字识别性能的重要方法之一,近年来受到了学术届的普遍关注。文章提出了一种基于单字单网的手写体汉字识别纯神经网络的多分类器集成方案,并通过实验证明该方案是行之有效的。
It is concerned commonly that the intuitionistic,compact method which combined some different classifiers to improve the recognition of HCCR.This paper proposes a scheme to HCCR,which is made up of pre_classifiers and model integrated,and both used neural networks.Experiments show that this method works and can classify large number of catalogs well.
出处
《计算机工程与应用》
CSCD
北大核心
2005年第28期44-45,87,共3页
Computer Engineering and Applications
基金
国家自然科学基金(编号:60275005)
广东省自然科学基金(编号:011611)
关键词
手写体汉字识别
多分类器集成
神经网络
handwritten Chinese character recognition,multi classifers integrated,neural network
