SINGLE IMAGE DEHAZING USING AIR LIGHT REFINEMENT
Publication Date : 14/03/2019
Conference Name :
Single image haze removal is important for many practical applications. But, dehazed results of the existing algorithm tend to be over smoothened with missing fine image details. This drawback is caused due to the following two factors: Inaccurate air light estimation and disregard of multiple scattering. In the proposed method, a dehazing algorithm which gives a detail preserving image is given based on two priors: depth edge aware prior and air light impact regularity prior. Based on the depth edge aware prior, an air light refinement algorithm is proposed. Here the gradient strength is employed to smoothen the channel. Based on air light impact regulatory prior, an adaptive sharpening model is established to enhance levels of detail. The proposed algorithm removes haze and enhances the level of detail in the image.
No. of Downloads :