Atmospheric Light Estimation Using Gradient Information
Li-dan Gu, Chuansheng Yang, Chao Wang · 2019
In this paper, an effective method of using gradient information to estimate atmospheric light is proposed, which contributes to handle the problem of low contrast and detail degradation in outdoor images taken under bad weather. This method based on the observation that the visibility of the region with dense fog in the image is low, which lead to small gradient value. We detect the dense foggy area of the image by cutting the expanding high gradient region in the corresponding place of the input image. Then, the global atmospheric light can be estimated directly. Experimental results on a variety of hazy images demonstrate the efficiency of the proposed method.