基于模糊方向特征的手写体汉字识别  

HANDWRITTEN CHINESE CHARACTER RECOGNITION BASED ON FUZZY DIRECTION FEATURE

作　　者：王正群[1] 叶晖[1] 孙兴华[1] 杨静宇[1] 

机构地区：[1]南京理工大学计算机科学与工程系,南京210094

出　　处：《模式识别与人工智能》2001年第3期317-320,共4页Pattern Recognition and Artificial Intelligence

摘　　要：方向特征在手写体汉字识别中得到了广泛应用,本文引入模糊数学的思想,提出了模糊方向特征,汉字的模糊方向特征的提取过程包括:(1)确定模糊网络;(2)抽取边缘点"横、竖、撇、捺"模糊方向属性特征;(3)合成模糊网络和边缘点模糊方向属性特征,生成汉字字符的模糊方向特征,我们应用模糊方向特征,实现了一个脱机非限制手写体汉字识别系统,对含有7600(19类)个汉字的测试集测试,取得了较好的识别效果。这表明,本文提出的模糊方向特征对手写体汉字的识别是有效的。Direction feature is widely used in handwritten Chinese character recognition. In this paper we propose a fuzzy direction feature based on fuzzy mathematics. The procedure of extracting fuzzy direction feature includes: (1) defining fuzzy mesh, (2) extracting the attribution feature of boundary points related to the four fuzzy directions-horizontal, vertical, left diagonal, and right diagonal, ( 3) combining fuzzy mesh with fuzzy direction attribution feature of boundary points to obtain the fuzzy direction feature of a Chinese character. Using fuzzy direction feature, we implemented an off-line recognition system for unconstrained handwritten Chinese characters. Experiments on a testing set containing 7600 Chinese charactes (19 classes) show that the fuzzy direction feature proposed in this paper is effective for handwritten Chinese character recognition.

关 键 词：手写体汉字识别 模糊方向特征 特征抽取 字符识别 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]