手写体字符识别的多特征多分类器设计  

Design of Multiple Features and Multiple Classifiers for Handwritten Character Recognition

作　　者：沈淑娟[1] 姜建国[1] 曹建春[1] 

机构地区：[1]西安电子科技大学计算机学院,西安710071

出　　处：《计算机工程与应用》2004年第16期116-118,共3页Computer Engineering and Applications

摘　　要：特征选取和分类器设计是字符识别系统设计的关键。文章针对手写体汉字和阿拉伯数字混和字符集的识别提出了依据不同的分类要求,分别选取不同的字符特征并采用神经网络多分类器进行识别的设计方法。实验结果表明,该方法用于手写体混合字符集的识别是行之有效的。The selection of feature sets and the design of classifiers are at the heart of a character recognition system.A novel method for the recognition of handwritten Chinese characters and digits is presented in this paper.Considering diverse classification requirements,different features are selected and an integration of multiple neural networks is uti-lized.The experimental results obtained demonstrate that the proposed approach is promising.

关 键 词：特征选取 神经网络 多分类器 手写体字符识别 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]