手写体汉字识别笔画提取方法的研究  

Stroke Extracting Method for Handwritten Chinese Character Recognition

作　　者：耿强[1] 马珏[1] 

机构地区：[1]江苏广播电视大学,江苏南京210036

出　　处：《江苏广播电视大学学报》2006年第6期41-43,81,共4页Journal of Jiangsu Radio & Television University

摘　　要：由于手写体汉字的复杂性,目前手写体汉字识别正确率不高。笔者提出在拐点提取方法上以像素跟踪法辅助最大距离法,在笔段合并中只考虑与交叉点相连的笔段,并提出具体算法,该方法可以有效克服拐点遗漏,笔段合并错误的弊端,使拐点的遗漏率大大降低,笔段合并正确率有较大改进,汉字识别正确率达到90%。An important recognition feature of handwritten Chinese characters is widely used in OCR systems. Handwritten Chinese characters are very complex and the recognition rate of Chinese characters is poor. This paper presents an appropriate method to do stroke analysis. Based on handwritten Chinese character recognition, the method improves the insufficiency of inflection-point abstracting and stroke connecting. Experiments show that the method is useful and the recognition rate of of handwritten Chinese characters is 90%.

关 键 词：手写体汉字识别 拐点提取 笔段合并 

分 类 号：H123[语言文字—汉语]