基于嵌入式Linux和OpenCV的车牌定位方法  

License plate location method based on embedded Linux and OpenCV

作　　者：田红鹏[1] 焦鑫[1] 

机构地区：[1]西安科技大学计算机科学与技术学院,陕西西安710054

出　　处：《计算机工程与设计》2014年第11期3908-3912,共5页Computer Engineering and Design

基　　金：科技部;财政部科技惠民计划基金项目(2012GS610101);陕西省自然科学基础研究计划基金项目(2012JQ8030)

摘　　要：针对光照不均匀图像中的车牌定位问题,提出一种基于嵌入式Linux和OpenCV的车牌定位方法。根据车牌特征,分别在HSV和RGB颜色空间对原始图像做直方图均衡化、二值化等预处理,得到两幅二值图像;根据字符跳变规律,在HSV颜色空间得到的二值图像中定位含有车牌的候选区域,在RGB颜色空间得到的二值图像中逐个提取候选区域并进一步确定车牌边界,从而提取车牌。实验结果表明,该方法能够对复杂背景和光照不均匀时的车牌图像做准确定位,在抗干扰性和灵活性两个方面性能均有显著提高,具有一定实用价值。Aiming at the vehicle license plate location of the uneven illumination images, a license plate location method based on embedded Linux and OpenCV was suggested. According to the characteristics of the license plate, the original image was prepro cessed （such as histogram equalization and binarization） in the RG]3 and the HSV color space to get two binary images, then the candidate regions that might contain the plate were extracted from the binary image of HSV color space according to the change between characters. In the binary image got from the RGt3 color space, the candidate regions were positioned one by one, and further the plate boundaries were determined, thereby the plate was located. The experimental results show that this method can locate the plate accurately in the images with complex backgrounds and uneven illumination. Compared to other methods, the ro- bustness and flexibility are improved by using the method suggested, and so it is practical in the traffic management.

关 键 词：车牌定位 嵌入式LINUX HSV颜色空间 数学形态学 最大熵 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]