Realistic Simulation of Underwater Scene for Image Enhancement

Abstract

In recent years, learning-based methods have performed remarkably well in underwater image enhancement (UIE), but their performance is limited by the lack of high-quality, diverse training datasets. Current underwater image datasets are unable to address the following three issues: intradomain gaps in underwater environments, interdomain gaps between synthetic and real data, and domain inaccuracies. To overcome these limitations, we construct a realistic underwater scene using 3-D graphics engine through a three-step approach: 1) integrate a simulation-specific underwater light propagation models to create volumetric fog; 2) employ physical model-based rendering for accurate light field simulation; and 3) configure scenes with parameters extracted from real underwater images. Based on this framework, we develop an UIE dataset [model-based underwater synthetic environment (MUSE)]. Experiments demonstrate that models trained on MUSE outperform those trained on conventional datasets, highlighting the effectiveness of our approach.