基于部件特征的分类方法及在汉字识别中的应用  

Character Classification Method Based on Component Character for Chinese Character Recognition

作　　者：李晓辉[1] 吴蓓[1] 董武[1] 陆发春[1] 

机构地区：[1]安徽大学,合肥230039

出　　处：《微电子学与计算机》2003年第10期17-19,25,共4页Microelectronics & Computer

基　　金：国家社会科学基金资助项目(01ETQ002)

摘　　要：文章针对汉字的特点,选取一定的轮廓加权值为部件特征向量的度量,提出了一种基于部件特征的特征提取方法。并根据部件的特征和属性,采用ID3算法训练出一个决策树。结合ID3决策树和模糊逻辑的概念,采用粗分类和精分类两级分类判别的方法,首先利用每个部件的权值将部件细分成可处理的小组,然后分析待识别的汉字字符属于哪个不可再分的小组,从而提高了汉字的识别速度和识别效率。A new character-pick-up method of is proposed in this paper,which is based on component,this new method aims at the characteristic of Chinese-Character,selects the certain contour gain as the measurement of component-character-vec-tor.Moreover based on the component-character and the com-ponent-property,a decision-making-tree has been trained by ID3algorithm.Joint the ID3decision-making-tree and the il-legibility-logic-concept,with coarse classifying and subdivision to distinguish which class the character is belonged to,firstly the component is classified to the group which can not be divid-ed further with the component gain,and then we put the Chi-nese Character we want to recognize into the corresponding class,thus the speed and the efficiency of recognition is en-hanced.

关 键 词：汉字识别 部件特征 分类方法 特征提取 汉字结构 计算机 汉字信息处理 ID3算法 决策树 

分 类 号：TP391.12[自动化与计算机技术—计算机应用技术]