Abstract The majority of existing physical model-based dehazing algorithms have problems, including color distortion and halo effects, when restoring hazy outdoor scenes that often contain large areas of the sky. This paper proposes a single image dehazing algorithm based on sky segmentation and an optimal transmission map to improve the quality of dehazed images containing sky regions. The proposed algorithm acquires the sky region of a hazy image by using the mean shift technique and prior information of sky color rules and estimates the atmospheric light by introducing adaptive threshold constraints based on the sky region. Next, a hazy image feature-based objective function is designed, and a transmission map is accurately estimated by introducing gradient domain-guided filtering. On this basis, images are restored with the atmospheric scattering model, and the final dehazed images are obtained through tone adjustment. The experimental results demonstrate that the proposed algorithm is robust and can effectively eliminate haze and enrich the edge details of images. Compared to other algorithms, the saturated pixel ratio of the present algorithm is approximately zero, indicating the preferable color saturation of the restored images.
展开▼
