An Image Dehazing Algorithm Based on Sky Region Detection
Baoqing Jiang, Xiaoyan Zhao · Proceedings of the 2nd International Conference on Computer Science and Application Engineering · 2018
Images1 obtained in foggy days are degraded because the scatting of the atmosphere. The scatting of the atmosphere makes the image blurred, lowers the color contrast and makes the object features hard to distinguish. To address the problem that dark channel prior out of work to the sky area of images and atmospheric light misjudgment. We propose the fog removal algorithm combined with sky region detection. Firstly, we convert input image into a greyscale image. From the greyscale image, we calculate its corresponding edge image with the canny operator. Then, we define a function of sky boundary position and the preliminary sky region estimated. We estimate atmospheric light in the detected sky area. Finally, we use a fusion of ideas to estimate the transmission map, which based on the luminance and dark channel prior model with the adaptive weights. Meanwhile, we also quote fast guided filter, which saves computation time and is faster. The results of study show that the method effectively removes fog and has good time efficiency compared to other algorithms. In terms of visual effect, the texture of the sky area well restored and the fogless image recovered real and natural.