Single Image Visibility Restoration Using Dark Channel Prior and Fuzzy Logic
Subhadeep Koley, Ahana Sadhu, Hiranmoy Roy, Soumyadip Dhar · 2018
Contrast, and colour-fidelity of images degraded by heavy fog should be restored to aid the ever emerging fields of computer vision applications. The revolutionary advent of self-driven cars has made it increasingly urgent that efficient and fast enhancement techniques be developed. Accumulation of dense fog is also a major contributor to road accidents. CCTV surveillance, object tracking, and Fog Vision Enhancement System (FVES) are other areas that can be facilitated by such algorithms. Standard filtering techniques that fail to effectively restore low contrast foggy images must be replaced with time-efficient special enhancement techniques. In this paper, we have used the Dark Channel Prior (DCP), which is the most renowned prior for fog removal. This prior assumes that a fog-free, clear image has intensity value close to zero in at least one colour channel. Although DCP effectively restores most of the colour information, it fails to enhance the contrast of the image. To overcome this shortcoming, we have proposed a Fuzzy Logic based Contrast Enhancement algorithm, which converts the image into the fuzzy domain and performs spatial operations on the image to restore the contrast adequately.