Haze Removal Algorithm Using Color Attenuation Prior and Guided Filter

In this paper, we describe the formatting guidelines for ACM SIG Proceedings. In this paper, we mainly introduce the haze removal algorithm which using color attenuation prior. We also propose a novel solution in the case of particular situations. Moreover, in order to make the algorithm more accurate, we propose a hierarchical searching method based on the quad-tree subdivision for estimating atmospheric light. In addition, we employ the shiftable window scheme to improve the guided filter, so that it can refine the depth map while effectively protecting image edge. Experimental result shows that our proposed method not only can obviously remove the haze from the haze image, but also the performance is more reliable in the case of particular situations.