Adaptive single image defogging based on sky segmentation

doi:10.1007/s11042-023-15381-2

2023-05-01

PRE

DOI：10.1007/s11042-023-15381-2

原文链接

原文求助

摘要

En 中文

Aimed at image distortion caused in process of defogging images and interfered by bright objects, a single adaptive image defogging method based on sky segmentation is proposed in this paper. Firstly, the combination of the dark channels and bright channels is used to estimate the atmospheric light. And then the sky region and the accurate atmospheric light are obtained by the information entropy and the dark channel prior. Next, the initial transmittance is estimated by the dark channel prior, and then the fast guiding filter is utilized to estimate the transmittance accurately. Further, an adaptive weight factor and an error amplification compensation factor of bright objects are introduced to optimize the transmittance mapping constraint and to correct the sky region and the non-sky region, respectively. Finally, bright adjustments are performed on the fog-free image obtained from the atmospheric scattering model by a nonlinear mapping method, which causes a clearer and more natural defogging image to be got. The experimental results show that adaptive single image defogging algorithms proposed in this paper has the higher value of peak signal-to-noise ratio, the higher value of structural similarity index measure, the higher value of naturalness image quality evaluator, the higher value of fog aware density evaluator, and the less time consumed, which illustrates that the adaptive single image defogging algorithm proposed in this paper has a better defogging effect on haze images with sky areas and bright objects.

Keyword:

Dark channel prior

Sky segmentation

Information entropy

Color distortion

Transmittance