An efficient dehazing method for edge enhancement by using entropy-map

Abstract

In this paper, we propose a novel dehazing method for effectively enhancing complex texture regions by using entropy information. The proposed method improves a drawback of an existing method due to estimating an incorrect transmission map. The proposed method consisted of two steps. In the first step, for effectively estimating a transmission map, an entropy map is generated from YCbCr color space. Then, by applying a threshold value to the entropy map, an input image is divided into complex texture regions and others. In the second step, in order to generate a transmission map, an averaging filter is applied to complex texture regions. And two edge filters are applied to non-texture regions. Consequently, a correct transmission map preserving more details in complex texture regions is generated. Experimental results have shown that the proposed method achieves better performance of image dehazing than existing methods.