基于改进ZS细化算法的手写体汉字骨架提取  

HANDWRITTEN CHINESE CHARACTER SKELETON EXTRACTION BASED ON IMPROVED ZS THINNING ALGORITHM

作　　者：常庆贺 吴敏华[1] 骆力明[1] Chang Qinghe;Wu Minhua;Luo Liming(Information Engineering College,Capital Normal University,Beijing 100048,China)

机构地区：[1]首都师范大学信息工程学院,北京100048

出　　处：《计算机应用与软件》2020年第7期107-113,164,共8页Computer Applications and Software

基　　金：国家自然科学基金项目(61672361);北京市教委-市自然基金联合项目(KZ201910028039)。

摘　　要：手写体汉字图像细化所提取出的骨架,突出了汉字的结构特征并减少了冗余信息,对手写体汉字的识别有着重要作用。Zhang-Suen细化算法迭代次数少、运行速度快,适合处理直线、T行交叉和拐角,但应用于细化手写体汉字图像时,细化后的汉字骨架无法保证单一像素宽,并且汉字骨架有毛刺。针对该问题,提出一种改进算法。使用消除模板和保留模板在保证手写体汉字骨架连续性的基础上,实现骨架的单一像素化;引进门限机制的判定方法,通过毛刺长度值与设定的阈值进行对比的方式去除了骨架毛刺。结果表明,改进算法实现了汉字骨架的单一像素化、无毛刺,准确突出了手写体汉字的拓扑结构。The skeleton extracted from the handwritten Chinese character image highlights the structural features of the Chinese characters and reduces the redundant information,which plays an important role in the recognition of handwritten Chinese characters.Zhang-Suen thinning algorithm has fewer iterations,fast running speed,and is suitable for dealing with straight lines,T-line intersections and corners,but when applied to thinning handwritten Chinese character images,the thinned Chinese character skeleton cannot ensure the width of a single pixel,and the skeleton has burrs.In order to solve this problem,we propose an improved algorithm.The elimination template and the retention template were used to ensure the continuity of the skeleton of handwritten Chinese characters,and the single pixelation of the skeleton was realized;we introduced the method of threshold mechanism,and removed the skeleton burr by comparing the length value of burr with the set threshold value.The experimental results show that the improved algorithm realizes the single pixel and no burr of the Chinese character,and accurately highlights the topological structure of handwritten Chinese characters.

关 键 词：细化算法 手写体汉字 冗余像素 毛刺 骨架 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]