基于模型匹配法的字符识别  

Character Recognition Basing On The Model Matching

作　　者：黄炯生 黄敏琪 

机构地区：[1]广东省电子技术学校,510515

出　　处：《中国科技信息》2008年第8期92-94,97,共4页China Science and Technology Information

摘　　要：本文采用模型匹配法对手写数字、印刷体数字和大写英文字母进行了识别,并采用Visual C++编程实现。其中,特征量的提取是关键。通过对图像的大小归一化、二值化、字符切割和细化等预处理之后,提取网格特征和交叉点特征两种特征量,并采用最短加权距离判别法进行匹配,从而获得较为精确的识别结果。The model matching is adopted to recognize the hand-written numbers, printed numbers and capitalization English letters in the paper. The Visual C＋＋＇s programming is used to implement it, In the recognition, the feature extraction is the key. After the character preprocessing such as the image normalization, binary, image cutting and thinning is done, two statistical features, that are grid feature and the cross featUre, are extracted. Then the recognition technique of the minimum weighting distance discriminance method is used. Thereby the recognition result is relatively exact.

关 键 词：模式识别 模型匹配法 手写数字识别 英文字母识别 数字形态学 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]