Advanced Dehazing of Single Satellite Imagery Using Enhanced Dark Channel Prior and Refined Transmission

Haze in photography reduces image contrast and light for distant subjects, with light dispersion by haze particles causing significant issues that affect image reliability in scenarios such as disaster assessments. Removing haze is crucial in various fields, including photography and computer vision applications. The Dark Channel Prior (DCP) method is commonly used for haze removal in single satellite images. Building on the DCP method, several enhancements have been proposed, such as the Multi-Scale Dehazing Network (MSDN), Color Attenuation Prior (CAP), CAP-Haze-Lines (CAP-HL), and DCP-Gamma Correction. This research focuses on integrating multiple techniques to mitigate haze effects in Synthetic Aperture Radar (SAR) imagery sourced from the Satehaze1k dataset. The proposed method begins with dark channel extraction, followed by adaptive atmospheric light estimation based on the brightest pixels in the dark channel. Transmission map estimation leverages these dark channels, and guided filters refine the transmission map. A Multi-Scale Dehazing technique is applied to address varying haze levels. Finally, the image reconstruction process utilizes atmospheric light and transmission maps, with additional RGB reduction and level shift techniques to minimize red color fluctuations. Integrating multiple dehazing techniques is crucial for superior image restoration in challenging haze conditions. The proposed approach not only enhances image quality but also balances visual improvement and processing time, making it a robust solution for single image dehazing in satellite imagery.