Emerging From Water: Underwater Image Color Correction Based on Weakly Supervised Color Transfer

By Chongyi Li, Jichang Guo, Chunle Guo. IEEE Signal Processing Letters, 2017

519 citations recorded by the source corpus.

Underwater vision suffers from severe effects due to selective attenuation and scattering when light propagates through water. Such degradation not only affects the quality of underwater images, but limits the ability of vision tasks. Different from existing methods that either ignore the wavelength dependence on the attenuation or assume a specific spectral profile, we tackle color distortion problem of underwater images from a new view. In this letter, we propose a weakly supervised color transfer method to correct color distortion. The proposed method relaxes the need for paired underwater images for training and allows the underwater images being taken in unknown locations. Inspired by cycle-consistent adversarial networks, we design a multiterm loss function including adversarial loss, cycle consistency loss, and structural similarity index measure loss, which makes the content and structure of the outputs same as the inputs, meanwhile the color is similar to the images that were taken without the water. Experiments on underwater images captured under diverse scenes show that our method produces visually pleasing results, even outperforms the state-of-the-art methods. Besides, our method can improve the performance of vision tasks.

Subfields: Computer Vision and Pattern Recognition, Media Technology

View Paper: https://www.semanticscholar.org/paper/26c303d214d58bcd8c6da252171f258590850fa2

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. Dacheng Tao
4. Wenqi Ren
5. Ming-Hsuan Yang
6. Runmin Cong
7. Saeed Anwar
8. Chen Change Loy

In Media Technology

1. Chongyi Li
2. Saeed Anwar
3. Runmin Cong
4. Junhui Hou
5. Peixian Zhuang
6. Chunle Guo
7. Wenqi Ren
8. Chunli Zhu

Citations in the corpus

• An Underwater Image Enhancement Benchmark Dataset and Beyond (2019, background): “From left to right are raw underwater images, and the results of fusion-based [31], retinex-based [33], histogram prior [45], blurriness-based [46], GDCP [40], Water CycleGAN [53], Dense GAN [55], the proposed Water-Net, and reference images.”
• U-Shape Transformer for Underwater Image Enhancement (2021, methodology): “For example, [28] directly used the CycleGAN [49] network structure, and [26] adopted a simple multi-scale convolutional network.”
• Underwater Image Enhancement via Medium Transmission-Guided Multi-Color Space Embedding (2021, methodology): “For unsupervised models [10], [17], they still inherit the disadvantage of GAN-based models, which produces unstable enhancement results.”
• PUGAN: Physical Model-Guided Underwater Image Enhancement Using GAN With Dual-Discriminators (2023, background): “…years, deep learning technology has also greatly promoted the development of the computer vision task, including object detection [26]–[33], medical analysis [34]–[37], remote sensing interpretation [38]–[41], content enhancement [13], [14], [42], [43], and underwater image enhancement [44]–[46].”
• Internet of Underwater Things and Big Marine Data Analytics—A Comprehensive Survey (2020, methodology): “Finally, the new concept of image reconstruction with DL [190], [250], [251] has promise for future research.”
• Underwater Single Image Color Restoration Using Haze-Lines and a New Quantitative Dataset (2018, background): “These metrics are somewhat heuristic and have limited applicability, as shown in [48].”
• Underwater Image Enhancement Using a Multiscale Dense Generative Adversarial Network (2020, methodology): “These competitive methods include FusionEnhance (FE) [3], RB [5], UDCP [7], CycleGAN [24], weakly supervised color transfer (WSCT) [21], and UGAN [17].”
• Underwater Image Enhancement With Hyper-Laplacian Reflectance Priors (2022, unknown): “Based on the cycle-consistent adversarial network (CycleGAN) [24], Li et al. [25] proposed a weakly supervised underwater color transfer model to relax the need of paired underwater images for training deep networks.”
• Zero-Reference Deep Curve Estimation for Low-Light Image Enhancement (2020, methodology): “CNN-based models usually use self-captured paired data for network training [5, 17, 28, 30, 32, 33] while GAN-based models elaborately select unpaired data [6,11,12,16,35].”