Fast image dehazing using improved dark channel prior

In the frog and haze climatic condition, the captured picture will become blurred and the color is partial gray and white, due to the effect of atmospheric scattering. This situation brings a great deal of inconvenience to the video surveillance system, so the study of defogging algorithm in this weather is of great importance. This paper deeply analyzes the physical process of imaging in foggy weather. After full study on the haze removal algorithm of single image over the last decade, we propose a fast haze removal algorithm which based on a fast bilateral filtering combined with dark colors prior. This algorithm starts with the atmospheric scattering model, derives a estimated transmission map by using dark channel prior, and then combines with grayscale to extract refined transmission map by using the fast bilateral filter. This algorithm has a fast execution speed and greatly improves the original algorithm which is more time-consuming. On this basis, we analyzed the reasons why the image is dim after the haze removal using dark channel prior, and proposed the improved transmission map formula. Experimental-results show that this algorithm is feasible which effectively restores the contrast and color of the scene, significantly improves the visual effects of the image. Those image with large area of sky usually prone to distortion when using the dark channel prior, Therefore we propose a method of weakening the sky region, aims to improve the adaptability of the algorithm.