纹理方向与边缘特性相结合的车牌定位方法  

Approach for License Plate Location Using Texture Direction and Edge Feature

作　　者：冯夫健[1] 谭棉[1] 王林[1] FENG Fu-jian TAN Mian WANG Lin(Guizhou Key Laboratory of Pattern Recognition and Intelligent System, Guiyang,Guizhou Minzu University, Guiyang,Guizhou550025, Chin)

机构地区：[1]贵州民族大学贵州省模式识别与智能系统重点实验室,贵州贵阳550025

出　　处：《贵州工程应用技术学院学报》2016年第6期139-144,共6页Journal of Guizhou University Of Engineering Science

基　　金：国家自然科学基金资助项目"道路车辆的信息检测及视觉理";项目编号:61263034;贵州省教育厅青年科技人才成长项目"基于视觉计算的人群异常行为监控预警系统研究";项目编号:黔教合KY字[2016]165

摘　　要：针对道路车辆的车牌信息提取问题,提出了一种基于纹理方向与边缘特性相结合的车牌定位方法。利用梯度方向场对获取原图像进行计算,从而得到纹理方向图;融合纹理方向图和二值图像通过区间判定边缘信息的方法确定候选区域及车牌区域;采用改进的区域标记方法实现车牌的准确定位。实验证明此种方法定位速度快、准确率高,具有很高的鲁棒性。This paper presents a method for license plate information extraction based on texture orientation and edge features under the road vehicles. The original image is calculated by the gradient field direction, thus obtain a texture pattern. This article adopts the method of interval decision edge information which combines the texture direction image and binary image to determine the candidate region and the license plate region. Finally, the plates are accurately positioned using the improved regional notation. Experimental results show that this method is positioning speed, high accuracy, with high robustness.

关 键 词：车牌区域 梯度矢量 纹理方向图 边缘特性 区间判定 

分 类 号：B84[哲学宗教—心理学]