摘要
采用改进型 CL AFIC(Class- Featuring Inform ation Compression)算法可以为学习子空间 L SM(L earningSubspace Method)算法提供更好的初始向量子空间 ,并通过 L SM算法对各类样本子空间按不同的旋转方式训练 ,来提高 OCR的识别率 .该文的特点在于首先采用了学习子空间算法来实现字符在灰度图像上的识别 ,它克服了传统的基于二值化图像进行特征提取和识别所带来的主要弊病 ,最大限度地保存了字符特征 .应用结果也表明 :采用改进型 CL AFIC的学习子空间算法 ,能在原有较高 OCR识别率的基础上得到进一步的提高 ,实用价值很高 .
The improved Class Featuring Information Compression (CLAFIC) algorithm can provide a better initial subspace for Learning Subspace Method (LSM). On the base of these subspaces, training of each subspace is rotated in different ways by LSM, which conduces to improving recognition rate of optical character recognition (OCR). The characteristic of this paper is to realize the optical character recognition by adopting LSM on character gray scale level, and therefore overcomes main shortages of classification on the binary scale level, and keeps integrity features of character information to the extreme. The results show that the effect of recognition has been improved by the CLAFIC LSM algorithm, which makes it highly worth applying to OCR fields.
出处
《计算机学报》
EI
CSCD
北大核心
2000年第7期679-684,共6页
Chinese Journal of Computers
关键词
改进型CLAFIC算法
学习子空间算法
字符识别
improved CLAFIC algorithm, LSM, recognition of gray scale Optical Character Recognition, image information features, recognition of similar character on shape
