首页 | 本学科首页   官方微博 | 高级检索  
     检索      

手写字符轮廓曲率的特征提取和识别
引用本文:王贵新,刘建胜,居琰,汪同庆.手写字符轮廓曲率的特征提取和识别[J].华中科技大学学报(自然科学版),2001,29(Z1):83-86.
作者姓名:王贵新  刘建胜  居琰  汪同庆
作者单位:1. 重庆后勤工程学院基础部
2. 重庆大学光电工程学院
摘    要:首先对手写体字符图像进行平滑,同时获取字符的轮廓点.然后提出了一个字符轮廓曲率计算的有效方法,为了利用小波变换对曲率信号进行去噪和平滑,该方法不是直接从轮廓点出发,而是进一步抽样得出新的轮廓点.最后得到了不受字符旋转、平移、大小、位置影响的字符特征,以及相应的识别算法.实验结果表明,它在单一的特征识别方法中优于其他方法.

关 键 词:手写体字符  曲率  特征提取  识别
文章编号:1000-8616(2001)1-0083-04
修稿时间:2000年10月24

Handwritten Character Recognition & Features Extraction from the Curvatures of Contour Points Based on Wavelet Transform
Wang Guixin Liu Jiansheng Ju Yan Wang Tongqing Doctoral Candidate., Dept. of Basics,Chongqing Institute of Logistics & Engineering.,Chongqing ,China ..Handwritten Character Recognition & Features Extraction from the Curvatures of Contour Points Based on Wavelet Transform[J].JOURNAL OF HUAZHONG UNIVERSITY OF SCIENCE AND TECHNOLOGY.NATURE SCIENCE,2001,29(Z1):83-86.
Authors:Wang Guixin Liu Jiansheng Ju Yan Wang Tongqing Doctoral Candidate  Dept of Basics  Chongqing Institute of Logistics & Engineering  Chongqing  China
Institution:Wang Guixin Liu Jiansheng Ju Yan Wang Tongqing Doctoral Candidate., Dept. of Basics,Chongqing Institute of Logistics & Engineering.,Chongqing 400011,China .
Abstract:The images of handwritten characters are smoothed and its contour points are obtained. In order to de noise and smooth the signal of the curvatures by the wavelet transform, an effective computer method of the curvatures of contour points is proposed which isn't directly based on the contour points but on the sampling. The character features that are unchanged by its rotating, parallel moving, size and its position transforming and corresponding recognition algorithm are got. Experimental results show that the method is superior to other methods while only one sort of features is adopted.
Keywords:handwritten character  curvature  feature extraction  recognition  
本文献已被 CNKI 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号