Image defogging algorithm based on dark channel improvement

• 1、School of electronic and information engineering, Liaoning Technical University, Huludao

Abstract：Hazy weather can cause serious damage to the image quality, which in turn affects the subsequent processing of the image, thus it is important to defog the fogged images. This thesis proposes an improved image defogging algorithm based on the dark channel to address the problems that the dark channel defogging algorithm is ineffective in processing sky areas, darkening the defogged images and failing to deal with colour distortion. Firstly, this algorithm is optimized by combining the atmospheric scattering model, basing on the original dark channel a priori theory. Secondly, a hierarchical search algorithm with quadtree subdivision is introduced to estimate atmospheric light values more accurately. Then, a gradient domain guided image filtering algorithm is used to further refine the transmittance, and finally, the defogged image is transformed from RGB space to HSV space to enhance the overall contrast by means of pixel magnification.Subjective and objective comparisons with classical defogging algorithms on synthetic and real fog datasets, and a large number of experiments show that the algorithm in this paper has significant advantages.

Keywords： dark channel a priori atmospheric scattering model quadtree search algorithm gradient domain guided image filtering