Video Dehazing Using Denoising Diffusion Probabilistic Model with Trilateral Filtering and Multi-Scale Boosted Dehazing Network

Outdoor videos captured in poor weather conditions, such as haze or fog, often suffer from reduced visibility, affecting both cinematography and surveillance applications. The light scattering and absorption caused by aerosols in the atmosphere and airlight reflection result in images with faded colors and decreased contrast. This paper presents a novel approach to video dehazing by leveraging Deep Learning (DL), specifically utilizing the Denoising Diffusion Probabilistic Model (DDPM) for haze removal and a Multi-Scale Boosted Dehazing Network (MSBDN) with Dense Feature Fusion for enhanced image quality. We refine the transmission map using trilateral filtering to achieve smooth edge transitions and improve dehazing performance. Our method is evaluated on both synthetic and real-world datasets, demonstrating its robustness and effectiveness compared to state-of-the-art dehazing algorithms.