Single Image Dehazing using a Weighted Fusion of Dark and Bright Channel Prior with Gamma Correction
Sudeep D. Thepade, Chaitanya M. Nawale, Mehul V. Suryavanshi, Chinmayee D. Taralkar, Ajinkya A. Patil · 2021 2nd International Conference for Emerging Technology (INCET) · 2021
Nowadays, haze or fog has become a major hurdle for several computer vision applications. The images captured under such scenarios usually have poor visibility. This is because haze primarily affects the air light and hides the scene details. Such images, when directly used for computer vision algorithms, affect the performance of these algorithms and cause unsatisfactory results. This paper proposes an improved haze removal technique that uses a prior-based approach for dehazing. The method makes use of a weighted fusion of Dark and Bright Channel Prior for estimating the transmission medium. Further, an improved version of Color Constancy Prior is used to calculate the ambient illumination and Gamma Correction was used for post-processing. The comparative analysis was done using the PSNR, SSIM, and Entropy scores have shown the efficiency of the proposed method.