A New Defogging Method with Nested Windows
Baojun Qi, Tao Wu, Hangen He · 2009
Images captured in foggy scenes suffer from degraded contrast and colors. It's imperative to overcome the weather effects and make outdoor vision systems more reliable. This paper describes a fast algorithm to enhance the visibility of foggy scenes with solely a single image. Firstly, the thickness of the fog is estimated with a Gaussian filter and an intensity image, which is composed of the foggy image's darkest color channel. It's based on the observation the edges of the airtight are similar to that of the foggy image's low frequency component. Secondly, to handle misestimates of the airtight, a revised imaging model and a strategy of nested multiple windows are proposed to enhance the visibility and make faint objects in the distance to be visible. Compared with previous single image based techniques, the algorithm proposed in this paper can handle images with large regions similar to the atmospheric light. In addition, it's timesaving. Experimental results on large number of foggy images validate the effectiveness of the method.