Single Foggy Image Restoration Using Orientation Extended Fields of Experts

Traditional single foggy image restoration algorithms cannot sufficiently describe the local directional structure information of the scene,which tends to make partial details of the scene be vague or missing after the recovery.To overcome this drawback,a new fog-removing algorithm based on orientation extended Fields of Experts was proposed in this paper.Firstly,the atmosphere transmission map was estimated cursorily according to the atmospheric scattering model,and it was further refined by the orientation extended Fields of Experts.To avoid the distortion at highlight areas,the atmosphere transmission map was restricted and adjusted by the no black pixel constraint algorithm.Secondly,the grads apriority rule was used to obtain the set of pixels that are located in the infinite region,so as to estimate the intensity of atmospheric light.Finally,by reversely solving the atmospheric scattering model,the defogged image was obtained.Experimental results show that the proposed algorithm can not only remarkably improve the detailed information of scene,and enhance the degree of clearness of foggy images,but also make the color appearances much more natural and realistic.