Abstract: The purpose of image defogging is to remove the fog effect from image of surveillance video to improve the fog haze image visual effect. Presently, there is only a comparison between images before and after defogging, and the results are often distorted seriously and oversatuarted. Thereby, it is hard to ensure the clear details and the integrity of color information simultaneously. For tackling above problems, a new optimized method for images recovering was proposed with combination of HIS luminance component and RGB space, which was based on atmosphere scattering model and optical principals. In this method the relative depth relationship of image scene was analyzed by comparing images in fine and haze days with help of the most eye-sensitive HSI luminance component. Finally, by utilizing atmosphere scattering model and the comparison of depth of field, the recovering and result evaluation were conducted on the video obtained in haze. The experimental results show that, compared with the defogging methods calculated in RGB space, the proposed method has more clear defogging results and smaller degree of color distortion and oversaturation.

Key words: video surveillance, defogging, atmosphere scattering model, image recovering, depth of field

摘要： 图像去雾技术处理的目的是消除雾霾对视频监控图像的影响,提高雾霾图像的视觉效果。目前,一般去雾图像只是比较去雾后和去雾前的图像,处理结果通常失真严重或过饱和,不能在保证细节清晰的同时保证颜色信息完整。针对上述问题,提出了一种基于大气散射模型和光学原理,建立具有散射特性的HIS亮度转换模型,并与RGB空间结合计算的图像复原方法。该方法通过分析晴天图像和雾霾图像的对比关系,结合HSI空间人眼视觉最敏感的亮度分量计算出图像场景的相对深度关系,利用大气散射模型以及景深比,对雾霾视频图像进行清晰复原和结果的测评。实验结果证明,与只从RGB空间计算的去雾霾方法对比,所提方法去雾效果更清晰,彩色失真和过饱和程度更小。

关键词: 视频监控, 去雾, 大气散射模型, 图像复原, 景深