融合多级光照处理的车牌图像二值化算法  

Binarization algorithm of license plate image based on multistage light processing

作　　者：马超玉 孙俊喜[1] 程帅[1] 阮光诗 刘红喜[1] 

机构地区：[1]长春理工大学电子信息工程学院,长春130022

出　　处：《计算机应用》2013年第A02期200-202,共3页journal of Computer Applications

基　　金：吉林省科技厅项目(2010312)

摘　　要：针对车牌识别系统中,车牌区域由于不均匀光照影响,使得车牌图像二值化不理想的问题,提出一种融合多级光照处理的车牌图像二值化算法。该算法首先对车牌图像进行顶帽变换降低光照对背景影响;然后用Retinex算法抑制光照对前景影响,使车牌字符亮度均匀;最后依Z字形逐行扫描图像,利用动态阈值进一步减少光照不均对二值化方法的影响。实验结果表明,该算法简单、实用,能较好地克服不均匀光照对车牌图像二值化的干扰,使二值化图像字符清晰。In order to solve the problem that the traditional binarization algorithm of license plate image is not ideal in the the license plate recognition system, due to non-uniform illumination conditions in the license plate region. This paper presented a binarization algorithm integrated with multistage light processing. Firstly, the algorithm used top-hat transformation to calibrate global shadow to eliminate the light in_quence of the background; and then, used the Retinex algorithm to let the license plate character brightness uniformity; finally, by Z shape progressive scanning for image, used a variable threshold value to further reduce the effect of uneven light on a binarization algorithm. The experimental results demonstrate that the algorithm is simple, effective, and can eliminate the effect of the uneven illumination on image binarization algorithm.

关 键 词：顶帽变换 RETINEX算法 动态阈值处理 不均匀光照 

分 类 号：TP371.4[自动化与计算机技术—计算机系统结构]