首页 | 本学科首页   官方微博 | 高级检索  
     

基于纹理及小波分析的车牌定位方法
引用本文:黄 卫,路小波,余彦翔,凌小静.基于纹理及小波分析的车牌定位方法[J].中国工程科学,2004,6(3):19-24.
作者姓名:黄 卫  路小波  余彦翔  凌小静
作者单位:东南大学教育部智能运输系统工程研究中心,南京,210096
摘    要:提出了一种基于纹理和小波分析的车牌定位方法。针对图像背景复杂,且车牌所占比例较小的特点,提出了一种确定基元分类阈值的二值化方法;根据车牌字符分布规律,提出了二值纹理基元分析方法,提取车牌候选区域;基于小波分析提取车牌区域竖笔画特征,采用隶属度定量表征车牌竖笔画特征、位置特征及形状特征,给出综合这些特征、从候选区域提取车牌区域的方法。测试结果表明,该方法正确定位率超过96%

关 键 词:纹理    小波分析    基元    车牌定位
文章编号:1009-1742(2004)03-0019-06
收稿时间:2003/10/8 0:00:00
修稿时间:2003年10月8日

A Method for License Plate Localization based on Texture and Wavelet Analysis
huangwei,luxiaobo,yuyanxiang and lingxiaojing.A Method for License Plate Localization based on Texture and Wavelet Analysis[J].Engineering Science,2004,6(3):19-24.
Authors:huangwei  luxiaobo  yuyanxiang and lingxiaojing
Abstract:This paper presents a method for vehicle license plate localization using texture and wavelet analysis. In view of the complex background and the small ratio of the license plate area to the whole picture, a method for finding a threshold for primitive classification is proposed. According to the distribution regularities of the characters in the license plate, a binary texture primitive analysis method is presented to extract the candidate license plate areas. A feature of the vertical strokes in the license plate is extracted using wavelet analysis. The feature of the vertical strokes, position feature and shape feature are quantificationally evaluated using membership degree, and a method for selecting the license plate area from the candidates by integrating all the features is described. A test shows that the proposed method offers a localization correctness rate of higher than 96%.
Keywords:texture  wavelet analysis  primitive  license plate localization
本文献已被 CNKI 维普 万方数据 等数据库收录!
点击此处可从《中国工程科学》浏览原始摘要信息
点击此处可从《中国工程科学》下载全文
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