Novel Single Image Dehazing Algorithm

Abstract

In order to improve the quality of haze degraded image, a novel method is proposed combining dark channel prior and the atmospheric degradation model. Firstly, adaptive block is performed to acquire the dark channel image; then, in HSV space, laplacian pyramid filter is required to obtain the low pass portion of V, further compensated by the single scale Retinex to achieve the final atmospheric light; moreover, for the smoothness trait of guided filter and its fast operation rate, it is suggested to gain the optimal transmission map. Experimental results show that this algorithm can effectively improve the visibility of the optical imaging system.