Video fog removal using Anisotropic Total Variation de-noising

Contrast and color of the acquired videos by a visual system are degraded under hazy and foggy weather, which will affect the real-world applications like surveillance system, unmanned vehicle and a driver assistance system. The video fog removal framework is the necessity of vision systems for outdoor real-world applications in bad weather to recover visibility. In this paper, a fog removal framework for videos has proposed for vision enhancement. The proposed framework excludes frame-wise procedure of defogging and adopts a collective approach for the estimation of atmospheric light and depth map estimation from all the video frames to minimize computational burden and to accelerate the process. Weighted Least Square (WLS) filtering and Anisotropic Total Variation (ATV) is implemeted for edge preserving and denoising, respectively. Qualitative and quantitative analysis is used for comparative study and observed the capability of sharp edge preserving. The proposed framework consists of limited number of popular quality parameters that are data driven and constants and remain same irrespective of the image under consideration.