Single image dehazing based on detail loss compensation and degradation

摘要

Most image dehazing methods are based on the dichromatic atmospheric scattering model. However, this model cannot explain blurring effects. In this paper, we present a detail loss compensating model, based on the analysis of multiple scattering model. By assuming that the foggy image has less detail than the image captured on a clear day, we use the Laplacian template and present a smoothing point spread function to estimate the detail loss of a foggy image. The fog noise is subsequently obtained based on the estimated detail loss, and the inherent radiance of the scene can therefore be restored. Experimental results show that our method have good dehazing effect and enhance the image sharpness to a certain extent.