Optimized Image Dehazing using Dark Channel Prior with Minimum Fusion and Improved Atmospheric Light Handling

Haze-induced image degradation in outdoor scenes, caused by atmospheric scattering and absorption, poses significant challenges for applications such as surveillance, autonomous navigation, and remote sensing. This research focuses on improving image dehazing by enhancing two key components: Dark Channel Prior (DCP) and Atmospheric Light Estimation (ALE). DCP exploits the statistical property that haze-free regions exhibit low minimum intensity values, aiding in accurate haze detection. ALE estimates atmospheric light from distant, less-scattered sources, ensuring precise haze removal. To further enhance dehazing performance, this study introduces a Minimum Fusion Technique, which selects the minimum intensity value for each pixel across multiple dehazed outputs, effectively preserving fine details and reducing artifacts. The proposed method is evaluated on the SOTS Outdoor dataset from the RESIDE benchmark, comprising 500 pairs of real-world hazy and clear images. By integrating optimized DCP, ALE, and the Minimum Fusion Technique, the approach achieves superior performance with a PSNR of 35.45 dB and an SSIM of 0.97, outperforming existing methods such as Scene-Specific DCP, TON, and FFA_NET. Additionally, it demonstrates reduced execution time and enhanced visual quality, making it a more efficient and reliable solution for real-world dehazing applications. These improvements establish the proposed method as a state-of-the-art approach for restoring clarity in outdoor images, contributing to more effective and accurate image reconstruction in challenging environments.