A Novel Wavelet-Based Image Defogging Using Dark Channel Prior and Guided Filter

Abstract

Recently, fog removal from images and videos has got tremendous importance in image and video processing for object detection, tracking and surveillance system, advance image editing, and many more which are poorly affected by the fog. A lot of works have been proposed so far for removing fog from image and video. It includes different methods like contrast enhancement, local color line model based and many more fog removal technique. In this paper, we propose a novel defogging method where guided filter based dark channel prior (DCP) is applied on low–low (LL) band of discrete wavelet transformation (DWT) coefficient of the intensity rectified image which is obtained after homomorphic filtering. To improve the sharpness of the output, unsharp masking (USM) is applied on high–low (HL) and low–high (LH) bands of the DWT coefficient. Finally, sharp and high contrast output is generated by applying contrast limited adaptive histogram equalization technique (CLAHE) to the inverse transformed image. The proposed method improves the overall quality of the defogged output image with respect to contemporary methods. The measured quality metrics of the proposed method are compared with some recent works. The quantitative and qualitative results confirm the claims.