基于单目视觉的车牌快速定位方法  

Rapid license plate location algorithm based on monocular vision

作　　者：杨先勇[1] 王会岩[1] 周晓莉[1] 刘东基 宋盼盼 

机构地区：[1]重庆大学,重庆400044 [2]山东文登师范学校,山东文登264400 [3]陕西渭河煤化工集团有限公司,陕西渭南714000

出　　处：《微型机与应用》2012年第15期45-47,共3页Microcomputer & Its Applications

摘　　要：提出了一种基于颜色空间理论和形态学结合的方法。首先对图像进行颜色空间转换,按要求提取出需要的特征颜色区域,再对颜色特征区域进行检测,对边缘进行灰度统计,按一定的比例进行筛选,并利用前方车辆位置变化的特点,对下一次车牌可能出现的区域进行粗定位,利于下一次更快速的定位。应用该算法对100幅车牌图像进行定位,定位准确率达90%,速度均在0.1s内。This paper presents a combination of theory and morphology of the color space. Firstly, image color space is conversed, and the required features of the color region detection to is extracted, Then, the color characteristics of the regional are tested, the gray level statistics on the edge is done, and is filtered by a certain percentage. By using characteristics of the changes of the vehicle location on the front, the coarse positioning of the possible license plate region for the next time is finished, which is benefit of a more rapid positioning for the next time. 100 license plate image are used, for positioning in this which is benefit of algorithm, the positioning accuracy rate is 90%, the speed is in 0.1 s.

关 键 词：颜色空间理论 形态学 车牌定位 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]