用于粘连手写体汉字切分的二次寻点算法  

The Segmentation of Handwritten Chinese Characters by Bi-extracting Points

作　　者：吴越[1] 赵宇明[1] 朱锴[1] 

机构地区：[1]上海交通大学图象处理与模式识别研究所,上海200030

出　　处：《计算机工程与应用》2004年第8期31-34,共4页Computer Engineering and Applications

基　　金：国家自然科学基金资助项目(编号:60075007)

摘　　要：离线手写体汉字的切分是离线汉字识别的基础,其中以粘连字体的切分最难。文章提出了一种对粘连手写体汉字粘连位置精确定位的算法。该算法包含两次寻点过程,第一步寻找波谷点,对粘连字的分割位置进行粗定位;第二步寻找理想切分点,即结合波谷点与波谷邻近特征点的综合位置,最终给出切分线定位,以实现粘连汉字的切分。The segmentation of the handwritten Chinese characters is the base of recognition,in which the division of the joining characters is the most difficult problem.A two-stage algorithm based on the structural geometric features of the handwriting is presented in this article.After an initial image processing,the candidates of trough point are picked out in the first step.A special method by extracting the crossing points and angel points of the strokes,using the trough point to help segment the joining characters and imitating the spanning stroke is presented during the second step.The second step is the crucial point of the whole algorithm and the special method of bi-extracting points is described in detail in this article.

关 键 词：汉字切分 数学形态学 二次寻点 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]