Image dehazing algorithm based on optimized dark channel and haze-line priors of adaptive sky segmentation

When dealing with outdoor hazy images, traditional image dehazing algorithms are often affected by the sky regions, resulting in appearing color distortions and detail loss in the restored image. Therefore, we proposed an optimized dark channel and haze-line priors method based on adaptive sky segmentation to improve the quality of dehazed images including sky areas. The proposed algorithm segmented the sky region of a hazy image by using the Gaussian fitting curve and prior information of sky color rules to calculate the adaptive threshold. Then, an optimized dark channel prior method was used to obtain the light distribution image of the sky region, and the haze-line prior method was utilized to calculate the transmission of the foreground region. Finally, a minimization function was designed to optimize the transmission, and the dehazed images were restored with the atmospheric scattering model. Experimental results demonstrated that the presented dehazing framework could preserve more details of the sky area as well as restore the color constancy of the image with better visual effects. Compared with other algorithms, the results of the proposed algorithm could achieve higher peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM) evaluation values and provide the restored image with subjective visual effects closer to the real scene.