基于MeanShift算法的西夏文字笔形识别  

Recognition of components in the Tangut script based on MeanShift algorithm

作　　者：孟一飞[1] 杨小花[1] 张晓彪[1] 

机构地区：[1]宁夏大学物理与电子电气学院,宁夏银川750000

出　　处：《广西大学学报（自然科学版）》2017年第3期1107-1113,共7页Journal of Guangxi University（Natural Science Edition）

基　　金：宁夏自然科学基金资助项目(NZ15048)

摘　　要：西夏文是辽宋西夏金时期西夏王朝使用的一种古文字。西夏文字研究是西夏历史人文研究的重要环节,而针对西夏字符的智能识别还是一个尚未被深入研究的新领域,提出基于西夏文四角编码规则,通过识别文字的四角构件确定文字的编码,实现对字符的识别。在识别文字构件的环节,提出采用Mean Shift核估计直方图统计方法,提取特征值,实现智能识别。Tangut character was used in Tangut Dynasty in the period of Liao Song Xixia Jin. The research of Tangut character plays a leading role in the historical and cultural research of Tangut.Intelligent recognition of Tangut character is still a field expecting for further research. It is suggested in the paper to recognize the Tangut character based on 4-corner code,and to implement the recognition by determining the code of components of 4 corners of Tangut character. In the process of identifying character components,Mean Shfit algorithm is adopted to extract the features of the corner components for intelligent recognition.

关 键 词：MeanShif't 西夏文 核函数 文字识别 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]