Underwater imaging plays a crucial role in deep-sea resource exploration, underwater inspection, and marine engineering. However, severe light absorption and scattering, uneven illumination, and complex water conditions often lead to low contrast, color distortion, and blurred details in captured images. These degradations significantly degrade the reliability of downstream underwater vision tasks, making underwater image enhancement and restoration essential yet challenging problems in underwater image processing. Although numerous enhancement and restoration approaches have been proposed, many existing methods still struggle with complex degradation patterns, limited generalization ability, and high computational cost. To provide a comprehensive overview of this rapidly evolving field, this paper presents a systematic review of recent advances in underwater image enhancement and restoration. First, the physical mechanisms of underwater imaging are analyzed, and representative underwater optical imaging models are introduced to explain the fundamental causes of image degradation. Based on these mechanisms, existing methods are systematically categorized, including traditional image processing techniques and recent deep learning-based approaches. Their underlying principles, architectural characteristics, advantages, and limitations are comparatively analyzed. Furthermore, widely used underwater image datasets and evaluation metrics are summarized to facilitate fair performance comparison. Representative methods are experimentally evaluated on public datasets through both qualitative and quantitative analyses, with particular attention to enhancement effectiveness. Finally, the major limitations of current approaches are discussed, and recent research progress is highlighted. Promising future research directions are also outlined to support the development of more robust and practical underwater vision systems.
🌊
