摘要
为解决复杂光线对车牌识别造成的困难,提出基于改进的Retinex理论的复杂光线图像增强算法.首先提取车牌图像的亮度分量,对亮度分量进行顶帽变换,该处理可以准确地提取车牌图像的背景,进而提取出车牌字符图像,有效地削弱复杂光线的影响;再利用改进的Retinex算子进行细节增强同时消除伪影,得到最终输出结果.实验结果表明,对于复杂光线下的车牌图像,文中算法在主观视觉效果和客观指标两方面都体现出充分的优势,证明了该算法的可行性.
In order to solve the problem caused by the complex illumination of license plate recognition, an image enhancement algorithm under complex illumination based on improved Retinex algorithm is presented. First of all, extract the brightness of the images and then process the brightness component with top-hat transformation, using this method the background of the license plate images can be extracted accurately and then extract the character, attain the purpose that effectively reduces the influence of the complex illumination; next enhance the image details and eliminate the halo at the same time with improved Retinex algorithm and get the final results. Experiments show that for the license plate images in complex light, the method in this paper fully reflects the advantages in both subjective visual effect and objective indicators, proves the feasibility of this method.
出处
《计算机辅助设计与图形学学报》
EI
CSCD
北大核心
2016年第6期976-985,共10页
Journal of Computer-Aided Design & Computer Graphics
