Single Image Atmospheric Veil Removal Using New Priors for Better Genericity

From an analysis of the priors used in state-of-the-art algorithms for single image defogging, a new prior is proposed to obtain better atmospheric veil removal. Our hypothesis based on physical model, considering that fog appears denser near horizon rather than close camera. It leads more restoration when depth important, natural rendering. For this purpose, Naka–Rushton function modulate according empirical observations synthetic foggy images. The parameters are set from features input image. This method also prevents over-restoration and thus preserves sky artifacts noises. algorithm generalizes different kinds fog, airborne particles, illumination conditions. extended nighttime underwater images by computing each color channel. Qualitative quantitative evaluations show benefit algorithm. evaluation shows efficiency four databases with types which demonstrates broad generalization allowed algorithm, contrast most currently available deep learning techniques.