Adaptive contrast enhancement for underwater image using imaging model guided variational framework

Underwater images are typically characterized by blurry details, poor contrast, and color distortions owing to absorption and scattering effects, which limits the performance of several high-level tasks. However, most existing approaches are incapable of removing these multiple corruptions elegantly. Hence, an imaging model guided variational framework is proposed to simultaneously address the corruptions. In this study, underwater imaging model is imposed on the variational framework to correct the deviated color. The differences of gray values in channel and space dimensions are proposed to maximize the contrast of enhanced images. Furthermore, an adaptive weight function is designed to address the issue of excessive enhancement. Finally, a coarse-to-fine strategy is employed to efficiently solving the variational framework. Owing to the reasonable framework, the proposed method can be well generalized to sandstorm images and hazy images. The provided experiments demonstrate that the proposed method presents the highest CIEDE2000, UIQM, and FDUM scores, i.e., 40.42, 0.82, and 5.03. These extensive experiments validate the superiority of proposed method in improving the quality of underwater images from both qualitative and quantitative perspectives.