Improved Single Haze Removal Algorithm Based on Color Attenuation Prior

Abstract

With continuous development and application of computer vision, images with high clarity and quality are increasingly important in training different kinds of models, especially in the field of automated driving and road safety monitoring. In this paper, through learning on the benchmarking dataset, we propose a novel quadratic model of color attenuation prior to calculate the scene depth map and then obtain the transmission matrix and restore the hazy-free image. Numerous experimental results show that the proposed approach is more accurate on sculpt the correlation between scene brightness, saturation and depth, therefore it outperforms the linear model in terms of both dehazing effect and clarity. Meanwhile, the objective and subjective comparison with the other state-of-the-art algorithms on the hazy images demonstrate the efficiency and effectiveness of the proposed method.