Images acquired under deprived weather environment are frequently corrupted\ndue to the presence of haze, mist, fog or other aerosols in a form of\nnoise. Haze elimination is essential in computer vision and computational\nphotography applications. Generally, there is the existence of numerous approaches\ntowards haze removal which are mostly meant for hazy images under\ndaytime environments. Although the potency of these proposed approaches\nhas been comprehensively established on daylight hazy images. However\nthese procedures inherit significant limitations on images influenced by\nnight-time hazy environments. Since night time haze removal dehazing remains\nan ill-posed problem, we proposed a novel method for night-time single\nimage dehazing which is efficient under night-time environments. The\nproposed scheme is a dark channel-based local image dehazing procedure\nthat locally estimates the atmospheric intensity for each selected mask on a\ncorrupted image independently and not the entire image. This is done in order\nto overcome the challenge of night-scenes that are exposed to multiple/\nartificial lights source and spatially non-uniform environmental illumination.\nWe performed an adaptive filtering on the combined dehazed masks\nto improve the degraded image. We validated the supremacy of the proposed\napproach in terms of speed and robustness through computer-based experiments.\nConclusively, we displayed comparison results with state-of-the-art\nand extensively emphasized the comparative advantage of our scheme.
Loading....