White balancing based improved nighttime image dehazing

The nighttime captured images mostly suffer from low contrast, color distortion, and glow effects due to the presence of haze (fog or smog), illumination variations, and artificial lighting. In this paper, we present an image enhancement framework, which successfully deals with the aforementioned effects faced while imaging at night time. At first, the proposed approach achieves glow decomposition by performing white balancing of the image. Next, we enhance the low-contrast areas by performing contrast-limited adaptive histogram equalization (CLAHE) which achieves image enhancement at the cost of introducing noise and edge distortion. In order to eliminate the introduced noise and preserve the edges we perform bilateral filtering. The proposed framework capitalizes on the individual strength of each method while encountering the artifacts produced during processing at each stage. We perform suitable tuning of the parameters of the methods involved for their performance coherence. The quantitative and qualitative comparison with the existing approaches demonstrate the increased effectiveness of the proposed approach in dehazing the nighttime images.