A Single Image Defogging Algorithm Based on Bright and Dark Region Segmentation

The defogging algorithms based on the dark channel prior have a poor effect in the scene with dim brightness or multiple bright and dark regions, and the defogged image is generally dark. In this paper, a single image defogging algorithm based on bright and dark region segmentation is proposed. Firstly, the foggy image is sharpened, and segmented into several super pixels by using the improved SLIC super pixel segmentation method. According to the brightness feature of the super pixels, they are classified into multiple bright and dark regions. Then, the transmittance and atmospheric light are calculated by using the bright and dark segmentation image and the physical model of atmospheric scattering, and the preliminary defogged image is obtained. Finally, HSV spatial brightness adjustment algorithm is used to correct the brightness of bright regions, and multi-scale Retinex algorithm with adaptive weights is used to enhance the brightness of dark regions to obtain the final defogged image. The experiment shows that the method in this paper can effectively avoid the influence of multiple bright and dark regions on estimation of transmittance and atmospheric light, and it has a good effect in ordinary scenes and scenes with dim brightness or multiple bright and dark regions, so it is more applicable than other defogging methods based on the dark channel prior. In addition, the defogging ability of the proposed method is comparable to that of defogging methods based on deep learning, and the real-time performance of the proposed method is higher than that of defogging methods based on deep learning.