Enhanced DCP filter for Real-World Hazy Scenes

Abstract

Haze is an atmospheric phenomenon that considerably degrades the visibility of out- door scenes. This happens due to atmosphere particles that absorb and disperse the sunshine. This paper introduces a unique single image visibility restoration algorithm that enhances visibility of such corrupted pictures. A unique edge-preserving decomposition-based technique is prepared to estimate transmission map for a haze image. Therefore, haze removal algorithmic rule has been taken from Koschmiedars law that includes a quick replacement-variation approach to dehaze and denoise at the same time. The proposed technique Enhanced DCP Filter (EDCPF) initially estimates a transmission map employing a windows adaptive technique that supported the dark channel. Restoration of foggy images is an important issue for the de-weathering in computer vision. A new method has been introduced for estimating the optical transmission in hazy scenes. Based on this estimation, the scattered light is eliminated to increase scene visibility and recover haze-free scenes.