摘要
汉字识别是目前识别领域中的难题,而细化是其预处理第一步。研究了基于主曲线骨架提取算法和基于模板的Hilditch算法两种手写汉字识别算法。通过程序实现两个算法分析比较了两种算法对于不同质量的汉字图像骨架提取效果。实验表明前者的鲁棒性能好,能适合各种质量差的图像,但是算法还有待于优化,而后者在图像质量差的情况下细化效果很差,但是算法速度很快。
Chinnese characters discemment is the hard problem in discernment field, and the thining is the first step to deal with. Studies two of the skeleton algorithm, one is based on principal curve, and the other is Hilditeh algorithm which is based on a template. Analysis of two different algorithms for the quality of the image of Chinese characters skeleton extraction. Experiments show that the former＇s performance well, and be able to fit a variety of poor quality images, but still to be optimization. Although the thining effect of the latter is poor,but the speed is fast.
出处
《计算机技术与发展》
2009年第7期114-116,共3页
Computer Technology and Development
基金
安徽省自然科学资助计划项目(2006KJ013A)
