基于车体对称及颜色聚类的车牌自动定位方法-计算机应用研究.PDF

第 11 期 杨文霞等: 基于车体对称及颜色聚类的车牌自动定位方法 2·25 · 基于车体对称及颜色聚类的车牌自动定位方法 * 1 1 2 杨文霞 , 黄樟灿 , 汪 斌 ( 1. 武汉理工大学 理学院 数学系, 湖北 武汉 430070; 2. 上海交通大学 图像处理与模式识别研究所, 上海 200030) 摘 要: 对车牌区域的准确定位是实现自动车牌识别的一个关键问题。提出了一种综合考虑车体对称特性以 及采用颜色聚类的高效车牌定位算法。该算法首先对待分析图像所属的场景进行识别, 自动将图像分类为白天 场景类或夜晚场景类; 然后针对不同场景图像使用不同的车牌定位方案进行处理。对夜间场景中的车牌采用二 次定位法, 而对白天场景中的车牌采用结合车体对称轴定位并考虑车牌区域文本特征的方法。实验结果显示, 该方法对图像场景分类以及车牌定位的总体准确率都达到了 95. 7% 。 关键词: 智能交通系统; 车牌定位; 车体对称; 颜色聚类 中图法分类号: TP391. 41 文献标识码 : A 文章编号: 1001- 3695( 2005) 11- 0225- 04 License Plate Location Based on Symmetry Finding and Color Clustering 1 1 2 YANG Wen-xia , HUANG Zhang-can , WANG Bin ( 1. Dept. of Mathematics, College of Science, Wuhan University of Technology, Wuhan Hubei 430070, China; 2 . Institute of Image Processing Pattern Recognition, Shanghai Jiaotong University, Shanghai 200030, China) Abstract: Accurate location of license plates in images plays an important role in automatic license plate recognition systems. A novel and effective method is proposed in this paper for automatic license plate location taking advantages of symmetry of the car body and employing color clustering. In this method, scene ( night or day) of the image to study is first identified. For each type of scene, a specific algorithm is designed to locate the license plate in it. A double-location algorithm is implemen- ted for night images, and another one based on symmetry finding and rules regarding text related characteristics in the region of license plate for day images. Experimental results show an accuracy of 95. 7% for both image scene classification and lice