An efficient single image dehazing algorithm based on transmission map estimation with image fusion

Abstract

Single image dehazing is always the focus of attention in the field of image processing. The major improvements in most optical scattering model based dehazing algorithms focused on the transmission map estimation since that the improper of transmission map estimation may cause over-high contrast or color distortion in the restored images. In this paper, we proposed an efficient single image dehazing algorithm in which the image fusion strategy is employed to adaptively improve the accuracy of transmission map estimation. Specifically, the transmission map obtained by the dark channel is corrected with Gamma transformation firstly. Then the original transmission map and its corrected version are fused with the weighting average image fusion algorithm to preserve the details in original transmission map. Moreover, the fused transmission map is filtered by guided filter to avoid the halo and block artifacts in the dehazed images. Finally, the dehazed images are reconstructed with the revised transmission map by optical scattering model. We test the proposed method on both natural and artifact haze images. The proposed method is compared with 8 state-of-the-art dehazing methods on three indexes include natural image quality evaluator (NIQE) index, structural similarity (SSIM), and peak signal-to-noise ratio (PSNR) indexes. Both subjective and objective evaluations demonstrate that the proposed method achieves competitive results.