Adaptive depth map-based retinex for image defogging

Abstract

Image defogging technology has attracted a lot of interest in the field of image processing. However, the structure characteristics of the fog images are rarely considered in the state-of-the-art defogging algorithms. To overcome this weakness, this paper proposes an adaptive retinex defogging method based on depth map for structure-complex fog images. First, based on the thickness of each scene, K-means algorithm is adopted to cluster image into several patches with similar structure characteristics. Then, for each patch, an adaptive single scale retinex model is built, which joints the mean depth of scenes in each patch and the retinex theory. Simulation results show that the proposed method offers comparable defogging performance to the conventional DCP and MSRCR methods, especially for the degraded images with a complex structure.