A Comprehensive Survey on Underwater Image Enhancement Based on Deep Learning

TL;DR

A detailed overview of the UIE task from several perspectives is provided, introducing the physical models, data construction processes, evaluation metrics, and loss functions, and identifying key areas for future research in UIE.

Abstract

Underwater image enhancement (UIE) presents a significant challenge within computer vision research. Despite the development of numerous UIE algorithms, a thorough and systematic review is still absent. To foster future advancements, we provide a detailed overview of the UIE task from several perspectives. Firstly, we introduce the physical models, data construction processes, evaluation metrics, and loss functions. Secondly, we categorize and discuss recent algorithms based on their contributions, considering six aspects: network architecture, learning strategy, learning stage, auxiliary tasks, domain perspective, and disentanglement fusion. Thirdly, due to the varying experimental setups in the existing literature, a comprehensive and unbiased comparison is currently unavailable. To address this, we perform both quantitative and qualitative evaluations of state-of-the-art algorithms across multiple benchmark datasets. Lastly, we identify key areas for future research in UIE. A collection of resources for UIE can be found at {https://github.com/YuZhao1999/UIE}.

Figures (10)

• Figure 1: Underwater degradation.
• Figure 2: Building feature maps with high-quality representation by window-based Transformer, Fourier Transformation and Wavelet Decomposition, respectively.
• Figure 3: Generating high-quality enhanced images with varying numbers of stages.
• Figure 4: Schematic diagrams of the Adversarial Learning, Rank Learning and Contrastive Learning in the UIE task.
• Figure 5: Schematic diagrams of improving the UIE model performance by different auxiliary tasks. The gray and green arrows represent forward and backward propagation, respectively.