DARK CHANNEL PRIOR WITH DETAIL ENHANCEMENT

Abstract

Atmospheric surroundings, mostly influenced by haze, obscure the scenes of outdoor photography and bring several kinds of troubles to computer vision and graphics applications. This becomes more problematic in applications like remote sensing, weather forecasting, etc. Hence, haze removal is a significantly and extensively accepted topic in the graphics field. Haze attenuates the light and blends with added features of the atmosphere. The main motive of “haze-removal” is recovering the image which is haze-free from such blended light. Proposed system is attempted to overcome the existing limitations like flickering artifacts, over-enhanced sky regions, high computational complexity and poor dehazing effects. The system uses a gradient filter-based Dark Channel Prior estimation (DCP) system for dehazing followed by a Detail Enhancement algorithm to further sharpen and highlight details. In detail enhancement, the image is smoothened out without affecting the edges. The amplitudes of the edges can be decreased but they are not blurred or sharpened. This prevents the halos and gradient reversals of the image.