Single Image Dehazing Using Dark Channel Fusion and Haze Density Weight

Abstract

Aiming at the defect that many traditional single image dehazing algorithms fail in sky areas and suffer from the darkness and residual artifacts, a method of single image dehazing based on dark channel fusion and haze density weight is proposed in this paper. Firstly, the method of dark channel fusion, which can effectively ensure that the pixel intensity of the sky areas in the dark channel map is as low as that of non-sky areas, is proposed to approximate the dark channel operation. Considering that the haze density is different in different regions of the hazy image, we then calculate the haze density weight to estimate reliable transmission. For obtaining more accurate transmission and atmospheric light, we utilize the guided filter to refine them. Consequently, the haze-free image can be restored by the atmospheric scattering model. Experimental results show that the proposed method can obviously eliminate the color cast and distortion phenomenon in sky regions and the details of restored image become more distinct after removing the residual artifacts. Moreover, the brightness, visibility and contrast are improved to a large degree.