Single image defogging

Abstract

Bad weather, such as fog and haze can significantly degrade the visibility of a scene. In order to overcome it, some approaches have been proposed. In this paper, we proposed a novel defogging algorithm just based on a single image. Firstly, we employ the method of dark channel prior as a foundation of our algorithm. After experimental analysis about the dark channel prior haze removal, we find that although dark channel prior behaves well in most situations, it also results in larger saturation values in some specific situations. Then, focusing on these situations, we propose an iterative algorithm to adjust the color distortion effected by higher saturation. Actually, this kind of global or local rectification can achieve a relatively ideal compromise between natural color and image definition. Experiments show that the algorithm proposed in this paper is good at dealing with the similarity-to-atmospheric-light objects defogging problem.