The link writing and font variation in Uighur text bring about certain difficulties to Uighur identification. In the analysis of the rules how Uighur words are constructed and, structural characteristics of the font,a standard character images template library can be established. And then match the collection of character images with the standard template. To those similar words among witch just have a slight difference to each other a secondary identification of the attachment image should be conducted to achieve a more accurate recognition of printing Uighur characters. Experiment recognition rate reaches to 94% .%维吾尔文字的连笔书写及字型变化的一些特征给识别带来一定的困难并会影响到识别的正确率.在分析了维吾尔文单词的组词规律及其字型结构特征基础上,采用一种基于区域分割模板匹配的识别方法,通过建立标准维吾尔文字母图像模板库,并与通过预处理所获得的待识别维吾尔文字母图像进行匹配.对一些相似度高且难区分的维吾尔文字母则采用提取这些相似字符的附属笔画部分的图像并对其按笔画的连通性、交叉性以及形态等特征进行附属笔画判定的方法来确定这些相似字符,从而较准确地实现了对维吾尔文印刷字符的识别.实验识别率达到94％.
展开▼
