Image Dehazing Based on Contrast Enhancement and Non-uniform Atmospheric Scattering Model

To overcome the limitations of the dark channel prior dehazing algorithm, such as color distortion and Halo artifacts, this study proposes an image dehazing method based on contrast enhancement and a non-uniform atmospheric scattering model. First, the hazy image undergoes adaptive histogram equalization with color preservation to enhance contrast and color saturation. Second, considering the non-uniform distribution of atmospheric light, a non-uniform atmospheric light estimation strategy is introduced. A coarse estimation is initially obtained and subsequently refined using a guided filter. The transmission map is further optimized using a weighted least squares filter to suppress block artifacts while preserving image details. Finally, the dehazed image is reconstructed using the non-uniform atmospheric scattering model. Both qualitative and quantitative experimental results show that the algorithm effectively removes haze, producing images with natural color representation and clear, detailed textures, outperforming existing methods.