基于小波隐性马尔可夫树模型的车牌定位  

Vehicle-License-Plate Location Based on Wavelet Hidden Markov Tree Model

作　　者：曹刚[1] 游志胜[2] 赵树龙[2] 

机构地区：[1]中兴通讯成都研究所,成都610041 [2]四川大学计算机学院,成都610064

出　　处：《电子科技大学学报》2004年第5期515-518,共4页Journal of University of Electronic Science and Technology of China

基　　金：国家自然科学基金资助项目(60272095);公安部科技基金资助项目(994454181)

摘　　要：根据小波隐性马尔可夫树模型对车牌纹理特征精确的多尺度描述,提出了一种基于WHMTM的车牌定位方法。该方法首先以小波域的高频滤波作为预处理,再根据车牌纹理的WHMTM参数采取从粗到精多尺度地分割车牌,从而快速、准确地在复杂的背景中定位车牌。实验结果表明该方法是一种速度快且抗扰性强的车牌定位方法。According to multiscale and accurate characterization of vehicle-license-plate (VLP) texture feature by wavelet hidden Markov tree model (WHMTM), a method of VLP Location based on WHMTM is proposed in this paper. At first high-pass filter in wavelet-domain is used as preprocessing. Afterwards VLP is segmented multiscaly from coarse to fine by WHMTM parameter of VLP. So VLP can be located from complex background fast and correctly. Experimental results show this method is very fast and robust.

关 键 词：车牌识别系统 车牌定位 小波变换 小波隐性马尔可夫树模型 

分 类 号：TN911.2[电子电信—通信与信息系统]