An Underwater Image Enhancement Benchmark Dataset and Beyond

By Chongyi Li, Chunle Guo, Wenqi Ren, Runmin Cong, Junhui Hou, Sam Kwong, Dacheng Tao. IEEE Transactions on Image Processing, 2019

2,021 citations recorded by the source corpus.

Underwater image enhancement has been attracting much attention due to its significance in marine engineering and aquatic robotics. Numerous underwater image enhancement algorithms have been proposed in the last few years. However, these algorithms are mainly evaluated using either synthetic datasets or few selected real-world images. It is thus unclear how these algorithms would perform on images acquired in the wild and how we could gauge the progress in the field. To bridge this gap, we present the first comprehensive perceptual study and analysis of underwater image enhancement using large-scale real-world images. In this paper, we construct an Underwater Image Enhancement Benchmark (UIEB) including 950 real-world underwater images, 890 of which have the corresponding reference images. We treat the rest 60 underwater images which cannot obtain satisfactory reference images as challenging data. Using this dataset, we conduct a comprehensive study of the state-of-the-art underwater image enhancement algorithms qualitatively and quantitatively. In addition, we propose an underwater image enhancement network (called Water-Net) trained on this benchmark as a baseline, which indicates the generalization of the proposed UIEB for training Convolutional Neural Networks (CNNs). The benchmark evaluations and the proposed Water-Net demonstrate the performance and limitations of state-of-the-art algorithms, which shed light on future research in underwater image enhancement. The dataset and code are available at https://li-chongyi.github.io/proj_benchmark.html.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/9496c94a6630a165a24db4e32f0b3f12a6616307

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

1. Chongyi Li
2. Chunle Guo
3. Saeed Anwar
4. Runmin Cong
5. Peixian Zhuang
6. Chunli Zhu
7. Liheng Bian
8. Lintao Peng

Citations in the corpus

• Underwater Image Enhancement With Hyper-Laplacian Reflectance Priors (2022, positive): “We first conduct comparisons on a large number of underwater images that contain various scenes and diverse degradations, where 300 underwater images are collected from [2], [3], [9], [13], [28], [39], [40].”
• Enhancement of Underwater Images With Statistical Model of Background Light and Optimization of Transmission Map (2019, background): “More recently, some researchers have began to work on underwater images databases [39], [40].”
• Using labeled data to evaluate change detectors in a multivariate streaming environment (2009, neutral): “All authors have read and agreed to the published version of the manuscript.”
• UIEC^2-Net: CNN-based Underwater Image Enhancement Using Two Color Space (2021, methodology): “Note that our test dataset and corresponding reference images are provided by UIEBD [1].”
• Underwater Image Enhancement via Weighted Wavelet Visual Perception Fusion (2024, methodology): “In addition, our WWPF has the highest overall mean quantitative score for the UCCS [14], UIQS [14], and UIEB [15] datasets.”
• Simultaneous Enhancement and Super-Resolution of Underwater Imagery for Improved Visual Perception (2020, methodology): “Figure 6: Qualitative comparison of Deep SESR-enhanced images with SOTA models: RGHS [29], UCM [33], MSFusion [4], MS-Retinex [72], Water-Net [43], UGAN [20], Fusion-GAN [44], and FUnIE-GAN [35].”
• U-Shape Transformer for Underwater Image Enhancement (2021, methodology): “Although UGAN [9] and Ucolor [26] could provide relatively good color appearance, they are often affected by local over-enhancement, and there are still some color casts in the result.”
• Underwater Image Enhancement via Medium Transmission-Guided Multi-Color Space Embedding (2021, methodology): “Unlike previous deep models [10], [17], [18] that only employ the features extracted from RGB color space, we examine the feature representations through a multi-color space encoder network, then highlight the most representative features via an attention mechanism.”
• PUGAN: Physical Model-Guided Underwater Image Enhancement Using GAN With Dual-Discriminators (2023, methodology): “Following the settings in [31], the training data include 800 pairs of underwater images selected from the UIEB dataset [2] and 1250 synthetic underwater images”
• Underwater scene prior inspired deep underwater image and video enhancement (2020, background): “[19] proosed a deep baseline model trained on the paired underwater imges and the corresponding reference images.”
• Learning to Enhance Low-Light Image via Zero-Reference Deep Curve Estimation (2021, background): “In addition, although deep neural networks have shown impressive performance on image enhancement and restoration [1], [2], [3], [4], [5], they inevitably lead to high memory footprint and long inference time due to massive parameter space.”