Source-linked AI summary
Underwater Optical Image Processing: A Comprehensive Review
Huimin Lu, Yujie Li, Yudong Zhang, Min Chen, Seiichi Serikawa, Hyoungseop Kim
TL;DR
Underwater cameras support ocean observation but produce images degraded by scattering, attenuation, and noise. This paper reviews state-of-the-art hardware and software processing methods across de-scattering, color restoration, and quality assessment, and summarizes future trends. It concludes by organizing approaches into imaging-type categories and analyzing representative methods and challenges.
Problem
Underwater optical images are degraded by water-dependent light transport and related conditions, creating processing challenges for cameras used in ocean observation.
Method
The paper comprehensively reviews underwater image-processing methods, organizing them into hardware- and software-based categories and covering de-scattering, color restoration, and quality assessment.
Results
The review discusses and analyzes state-of-the-art approaches across the two imaging classes and summarizes methods and future trends.
Takeaways & Limitations
The paper provides a structured overview of underwater image-processing approaches and identifies future trends in designing and processing underwater imaging sensors.
Takeaways & Limitations
Laser imaging instruments are seldom used in industrial applications because environmental susceptibility and complex device settings constrain their use.
Abstract
from arXiv · showhide
Underwater cameras are widely used to observe the sea floor. They are usually included in autonomous underwater vehicles, unmanned underwater vehicles, and in situ ocean sensor networks. Despite being an important sensor for monitoring underwater scenes, there exist many issues with recent underwater camera sensors. Because of lights transportation characteristics in water and the biological activity at the sea floor, the acquired underwater images often suffer from scatters and large amounts of noise. Over the last five years, many methods have been proposed to overcome traditional underwater imaging problems. This paper aims to review the state-of-the-art techniques in underwater image processing by highlighting the contributions and challenges presented in over 40 papers. We present an overview of various underwater image processing approaches, such as underwater image descattering, underwater image color restoration, and underwater image quality assessments. Finally, we summarize the future trends and challenges in designing and processing underwater imaging sensors.
1. Introduction
Underwater optical imaging is constrained by water-dependent light transport, artificial illumination, and biological or suspended matter, producing scattering, color loss, noise, and uneven lighting. The paper reviews hardware and software de-scattering, color restoration, quality assessment, and future directions.
- Scattering blurs underwater photographs, while wavelength absorption reduces captured-image color and sediments affect high-dimensional imaging.
- Artificial lighting produces vignetting through non-uniform illumination, while sunlight flicker creates strong highlights in shallow-ocean images.
- Underwater optical imaging is challenging because light attenuation, scattering, non-uniform lighting, shadows, color shading, suspended particles, and marine life affect captured images.
- The review covers hardware-based and software-based underwater image de-scattering methods.
- It also summarizes four color-restoration methods, reference-based and non-reference quality indexes, and future research trends.
2. Types of Underwater Imaging
Underwater imaging methods are organized into hardware- and software-based approaches, with hardware methods exploiting polarization, gated light, fluorescence, or stereo information. These approaches can improve visibility, but laser systems remain constrained by environmental sensitivity and complex device settings.
- Underwater imaging methods are categorized into hardware-based and software-based approaches.
- The four traditional hardware approaches are polarization, range-gated imaging, fluorescence imaging, and stereo imaging.
- Polarization imaging: Polarization imaging reduces backscatter by using polarization information from filters or polarized illumination.
- Range-gated imaging: Range-gated imaging selects reflected object light while blocking backscatter with a flash gate in turbid water.
- Limitations: Laser-imaging methods are susceptible to environmental conditions and complex device settings, so instruments are seldom used in industrial applications.
- Fluorescence imaging: Fluorescence imaging can recover underwater-scene shape and detect microorganisms in coral reefs.
- Stereo imaging: Stereo imaging estimates visibility coefficients using real-time algorithms applied to autonomous underwater vehicles.
3. Underwater Image Processing
The review organizes underwater image processing around physical and non-physical de-scattering, color restoration, and methods addressing scattering, haze, noise, and color distortion.
- 3.1 De-scattering: Physical model-based methods estimate turbidity, depth, wavelength compensation, or attenuation to recover clearer underwater images.Approaches include color-lines with Markov Random Fields, dark channel priors with guided filtering, and wavelength compensation.
- 3.1 De-scattering: The dark channel prior with guided filtering can achieve real-time processing.
- 3.1 De-scattering: Non-physical methods include local histogram equalization, CLAHE, exposure fusion, red-channel restoration, variable-kernel filtering, and frequency-domain enhancement.
- 3.1 De-scattering: Several methods retain limitations, including poor performance in very dark environments, residual non-uniform lighting, inadequate high-turbidity scatter removal, halos, noise, and color distortion.The cited methods respectively report these limitations across local equalization, local-region processing, exposure fusion, variable-kernel filtering, depth refinement, and color enhancement.
- 3.2 Underwater Image Color Restoration: Color restoration methods model light absorption and camera spectral response, estimate pixel colors, compute attenuation coefficients, or invert attenuation after color-space processing.Representative approaches use Markov Random Fields, hyperspectral imaging, mathematical stability models, quaternions, and spectral-response modeling.
4. Underwater Image Quality Assessment
The review treats underwater image quality assessment as important for measuring enhancement performance and surveys full-reference, no-reference, perceptual, and underwater-specific measures.
- 4. Underwater Image Quality Assessment: Underwater image quality assessment is important for measuring the performance of different image-processing methods.
- 4. Underwater Image Quality Assessment: Proposed measures evaluate structural distortion, colorfulness, sharpness, contrast, visibility of artifacts, overall quality, chroma, saturation, or contrast.
- 4. Underwater Image Quality Assessment: UIQM combines colorfulness, sharpness, and contrast measures for underwater images.
- 4. Underwater Image Quality Assessment: UCIQE uses a linear combination of chroma, saturation, and contrast in the CIELab color space.
- 4. Underwater Image Quality Assessment: Mean angular error assesses robustness and behavior with respect to underwater noises, while Qu evaluates similarity of image structures and colors.
- 4. Underwater Image Quality Assessment: Future challenges include non-uniform artificial lighting, inhomogeneous de-scattering, high-turbidity reconstruction, image reflection, computational imaging, deep learning, cloud computing, and the Internet of Things.
5. Conclusions and Future Trends
The paper comprehensively reviews underwater image processing by organizing methods into imaging-based categories and examining state-of-the-art approaches. It discusses wavelength compensation, physical and non-physical models, color reconstruction, quality assessment, and future trends.
- The review divides underwater image processing methods into two categories according to their imaging types.
- State-of-the-art approaches in both classes are discussed and analyzed in detail.
- Software-based processing includes wavelength compensation approaches using physical and non-physical models, as well as color reconstruction.
- The review also summarizes underwater image quality assessment methods and future trends.