A Framework for Outdoor RGB Image Enhancement and Dehazing

A framework for image visibility restoration and haze removal is proposed. The proposed technique utilizes hybrid median filtering in conjunction with accelerated local Laplacian filtering for initial dehazing of images. For visual enhancement and correct restoration of colors, constrained l(0)-based gradient image decomposition is applied. The proposed technique not only effectively removes haze from the images but also addresses the issues of distorted colors, visual, and halo artifacts, and haze removal from sky region in images in a better way when compared to other techniques. Experiments were performed on outdoor RGB images as well as remotely sensed images. The effectiveness of our proposed technique is demonstrated by quantitative and visual analyzes.