印刷体汉字识别方法的探讨

印刷体汉字识别方法的探讨 THE INVERSTIAGTION OF RECOGNITION METHODS FOR PRINTED CHINESE CHARACTERS 黄俊杰 Huang Junjie first-author <正> （一） 引言汉字的自动识别是模式识别领域中有重要意义的课题之一。国外已有过一些研究,尤其是近些年来,国外学者在汉字识别方面相继做了相当数量的工作。主要分为三种:印刷汉字的识别,手写体汉字的识别,在线手写汉字的识别（利用笔点运动信息的所谓在线实时识别系统）。而大量的工作都集中于印刷体汉字的识别。 1966年识别印刷体汉字的工作R.Casey和G.Nagy的报导[1],他们使用了两阶段的最简单的模板匹配技术。先采用64个所谓"组模"用来把汉字集合按相类的柘朴结构特征分成不同的字组。然后在字组中分划出单个模元进行最后识别。在1973年前后P.Wang In this Paper we discribe a mathematical model to recognize the printed chiuese elaraters, and discuss the design of the optical character recognition sgstem by meaus of the strature of sacr charaters. Llsing the idea of curuilinear integrcel for information compression, we have proposed a two stage recognition method with the combination of hardware and software, and the followiug cocepts have been introduced. Let the root-of-character G be formed by anumber of simple clased curves G 1 , G 2 , …, G n . Then we have the following property! where the curuilinear integrcls are tho second kine taking on the contors G 1 , G 2 , …, G n . Using these concepts we have found a desision method to recognize the printeed chinese chavaters, and made some extensions. 1980-04-01 2021-04-01 4