论文部分内容阅读
本文对70年以来,获得一定实验结果的“印刷体汉字识别方法”,进行扼要的介绍和评论。文中叙述了:分级图形匹配法,轮廓投影法,对判定加权相关法以及有阶段结果的变换“算法”和已取得较好结果的复合类似度法等。同时还简介了汉字识别中一些予处理技术,如文字线幅度的检测校正,文字位置的校正,文字图形的模糊化等。本文还提出了对开展“汉字识别”的一些不成熟的看法。
This article gives a brief introduction and comment on the “Printed Chinese Character Recognition Method” which has obtained some experimental results in 70 years. The article describes: grading graphics matching method, contour projection method, to determine the weighted correlation method as well as the transformation of the stage results “algorithm” and has achieved good results composite similarity method. At the same time, some preprocessing techniques in Chinese character recognition are introduced, such as the detection and correction of the width of the word line, the correction of the position of the text and the blurring of the text and graphics. This article also puts forward some immature opinions on carrying out “Chinese character recognition”.