DEEP LEARNING BASED ON MULTISTAGE PROCESS FOR IMAGE REFINING SYSTEM

Rain, Adverse weather conditions like rain, fog, and haze significantly degrade outdoor image quality, affectingvisibility and image detail restoration. Traditional image enhancement techniques struggle with these challengesdue to poor pixel restoration and high noise. This paper presents a multistage deep learning-based image refiningsystem that leverages Dark Channel Prior (DCP), Transmission Map Refinement via Deep Learning, AtmosphericLight Estimation, and Guided Image Filtering (GIF) to enhance visibility in affected images. The system improvesdriver assistance, surveillance, and traffic management efficiency by enabling clearer image outputs in real-time.