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.
出处
《模式识别与人工智能》
EI
CSCD
北大核心
2001年第3期317-320,共4页
Pattern Recognition and Artificial Intelligence
关键词
手写体汉字识别
模糊方向特征
特征抽取
字符识别
Handwritten Chinese Character Recognition, Fuzzy Direction Feature, Feature Extraction
