基于结构知识的手写体汉字部件提取算法  

Extraction algorithm of handwritten Chinese character components based on structure knowledge

作　　者：李泽瑶 李成城 LI Ze-yao;LI Cheng-cheng(College of Computer Science and Technology,Inner Mongolia Normal University,Hohhot 010022,China)

机构地区：[1]内蒙古师范大学计算机科学技术学院,内蒙古呼和浩特010022

出　　处：《计算机工程与设计》2023年第5期1479-1486,共8页Computer Engineering and Design

基　　金：内蒙古自治区科技计划基金项目(2020GG0167)。

摘　　要：针对手写体汉字结构复杂、部件粘连等问题,提出一种基于结构知识的手写体汉字部件提取方法。依据汉字结构特征,结合投影法制定连通区域合并规则,实现汉字部件初步提取。在此基础上,针对部件间存在粘连等书写不规范问题,对传统滴水算法进行改进,实现粘连部件分割。经实验验证,该方法较传统滴水算法在粘连部件切分方面的效果有明显提升,可实现对多种结构汉字的部件提取,具有较高的实用价值。To solve the problems of complex structure and adhesive components of handwritten Chinese characters,a method of component extraction of handwritten Chinese characters based on structural knowledge was proposed.Based on the structure characteristics of Chinese characters and projection method,the connected region merging rules were formulated to realize the initial extraction of Chinese character components.On this basis,the traditional dripping algorithm was improved to realize the segmentation of the adhering components in view of the non-standard writing problems such as adhesion between components.Experimental results show that the proposed method is better than the traditional dripping algorithm in the adhesion components segmentation,and can realize the extraction of various structural Chinese characters components,which has high practical value.

关 键 词：手写体汉字 汉字结构 投影法 连通区域合并 粘连部件 滴水算法改进 汉字部件提取 

分 类 号：TP391.43[自动化与计算机技术—计算机应用技术]