针对交通视频监控系统在恶劣雾霾天气下获取的图片质量受损，无法正确识别车辆信息的 问题，首先总结了已有图像除雾技术的研究成果，对比分析了全局直方图去雾处理法、同态滤波 去雾处理法、多尺度Retinex(MSR)去雾处理法3 种算法。然后，采用这3 种方法对雾霾交通图片 进行处理，并根据图像纹理特征统计量对处理后的图像效果进行了客观评价。仿真结果表明，3 种算法中，采用多尺度Retinex(MSR)处理法最能提高雾霾图像的质量，能有效地恢复和增强图像 信息，使图像的近景和远景得到均衡。最后，基于Matlab 设计了GUI图像去雾系统，逐一验证了 3 种算法的处理效果，结果表明该系统能够有效地提高交通视频监控系统图像的清晰度，可在一 定程度上优化雾霾天气下的交通图像质量。

Due to the badly hazy weather, it becomes more and more difficult for traffic supervision sys⁃ tem to collect high quality urban traffic images. Firstly, some defogging algorithms were summarized and studied, and three kinds of defogging methods, which were Global Histogram, Homomorphism Filter, and Multi Scale Retinex (MSR), were contrasted. Secondly, the traffic foggy image was processed by these three methods and the processed effect was evaluated according to the texture features. The result shows that MSR algorithm is the best method for improving the image quality, which can restore and enhance image information effectively, as well as balance the display of close and far scene of image. Finally, an image processing operation system of GUI based on MATLAB was designed. The effectiveness of these three algorithms was verified by the system one by one and it is concluded that the system can effectively improve the clarity of image and optimize the traffic foggy images.