基于支持向量机的联机手写维吾尔字符识别

SVM-BASED RECOGNITION ALGORITHM OF ONLINE HANDWRITTEN UYGHUR CHARACTERS

提出一种基于结构特征的手写维吾尔字符识别算法,首先根据字符的笔画数目将待识别字符划分为五个子集,然后再根据"附加笔画位置"等特征对字符集再进行划分。根据每个子集中的字符分布情况,提取不同长度的特征向量,然后利用SVM为每个字符集构造一个分类器,进行训练和识别。

This paper presents a recognition algorithm of handwritten Uyghur characters based on structural features.According to the number of strokes,the characters are first divided into five subsets,and then each subset is further classified on the basis of features such as ＂the position of additional strokes＂ and so on.Based on characters distribution in each subgroup,feature vectors with different-length are extracted,and then for each character set a classifier is constructed using SVM in order to train and recognise.