SVM在小字符集手写体汉字识别中的应用研究  

Small-set Off-line Handwritten Chinese Characters Recognition Based on Support Vector Machine

作　　者：朱辉[1] 杨扬[1] 颉斌[1] 封筠[1] 

机构地区：[1]北京科技大学信息工程学院,100083

出　　处：《微计算机信息》2004年第4期74-75,88,共3页Control & Automation

摘　　要：本文将支持向量机(SVM)引入到小字符集脱机手写体汉字识别中。文章首先介绍了SVM的基本原理和主要算法,然后在实验中采用了LibSVM训练软件,针对银行票据手写汉字的小字符集进行了仿真,同时与欧氏距离分类方法进行了比较。实验结果表明此方法的汉字识别率较高,在小字符集手写体识别中具有较强的实用性。This paper presents the application of SVM in small-set off-line handwritten Chinese characters recognition.The paper begins with basic introduction of SVM theories and algorithms. Then software LibSVM is pro -posed for handwritten Chinese characters training of bank forms. The results are also compared with Euclidean Distance classifier, which indicates that the SVM strategy can mprove recognition rate and therefore has more practicability.

关 键 词：SVM 支持向量机 小字符集 手写体汉字 汉字识别 LIBSVM 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]