Towards foggy image optimization: dark channel prior via RGB-splitted processing

The dark channel prior algorithm can reduce the impact of fog from an image, which facilitates a number of different applications such as target detection and target tracking. However, the current dark channel prior algorithm fails to solve the problems such as color restoration and image detail, that is algorithm faces a bottleneck. To solve this, we propose an adaptive color algorithm based on the dark channel prior(ACDCP), where the image is split into R,G,B channels, and then Gaussian filtering is performed on each channel. the transmittance of each channel is taken into account to correct the actual value of the atmospheric light value by the offset coefficient C. We qualify visual perception and quantify index analysis value, involving Laplacian gradient, Brenner gradient, SMD, and Vollath function value. The results show that the proposed ACDCP algorithm can effectively improve the color of the image restoration, and achieve a pronounced defogging effect.