Robust Sandstorm Image Restoration via Adaptive Color Correction and Saturation Line Prior-Based Dust Removal

Shan Zhou, Fei Shi, Zhenhong Jia, Guoqiang Wang, Jian Huang

Abstract: Enhancing the visibility of outdoor images under sandstorm conditions remains a significant challenge in computer vision due to the complex atmospheric interference caused by dust particles. While existing image enhancement algorithms perform well in mild sandstorm scenarios, they often struggle to produce satisfactory results in more severe conditions, where residual color casts and chromatic artifacts become pronounced. These limitations highlight the need for a more robust and adaptable restoration method. In this study, we propose an advanced algorithm designed to restore sand-dust images under varying sandstorm intensities, effectively addressing the aforementioned challenges. The approach begins with a color correction step, achieved through channel compensation and color transfer techniques, which leverage the unique statistical properties of sand-dust images. To further refine the restoration, we improve the boundary constraints of the saturation line prior (SLP) by adjusting the local illumination in the atmospheric light map, enhancing the model’s robustness to environmental variations. Finally, the atmospheric scattering model is employed for comprehensive image restoration, ensuring that color correction and dust removal are optimized. Extensive experiments on real-world sandstorm images demonstrate that the proposed method performs on par with state-of-the-art (SOTA) techniques in weaker sandstorm scenarios, showing marked improvements in more severe conditions. These results highlight the potential of our approach for practical applications in outdoor image enhancement under challenging environmental conditions.

External IDs:doi:10.3390/app15052594