A Semi-Supervised Domain-Adaptive Framework for Real-World Underwater Image Enhancement

Abstract

Underwater optical remote sensing is crucial for geoscience applications but often suffers from image degradation due to complex underwater environments. While learning-based methods have advanced underwater image enhancement (UIE), their efficacy in real-world UIE applications still faces challenges. This limitation arises from training predominantly on synthetic underwater images, resulting in a significant interdomain gap when applied to real-world data. In addition, diverse underwater conditions introduce intradomain challenges, such as color casts and haze, further complicating the UIE process. To address these issues, we propose SSD-UIE, a semisupervised domain-adaptive framework designed to mitigate both interdomain and intradomain gaps. Our approach employs a systematic synthesis pipeline to reduce visual interdomain discrepancies and introduces a large synthetic-real underwater image dataset (LSRUID) to facilitate the training of the framework. The semantic blender is developed to handle semantic interdomain differences, while the intradomain-aware feature extraction (IFE) branch and the feature alignment strategy effectively address intradomain variability. Furthermore, the Dual-Trans Block is introduced to enhance the UIE performance while maintaining computational efficiency. Extensive experiments demonstrate that SSD-UIE outperforms state-of-the-art (SOTA) UIE methods in both qualitative and quantitative evaluations on real-world underwater images. The codes and dataset will be publicly available at https://github.com/RockWenJJ/SSD-UIE.git