Pith. sign in

REVIEW 4 major objections 6 minor 40 references

OceanLens: An Adaptive Backscatter and Edge Correction using Deep Learning Model for Enhanced Underwater Imaging

T0 review · 4 major / 6 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read OceanLens claims that splitting an underwater image into estimated backscatter and attenuation, with Sobel and LoG edge losses, cuts GPMAE by 65% and raises UIQM by 60% over SeeThru and DeepSeeColor.

desk verdict A modest loss-function extension of DeepSeeColor whose headline numbers do not survive contact with its own tables. read the letter →

arxiv 2411.13230 v1 pith:KT3OXIVF submitted 2024-11-20 eess.IV

Rajini Makam , Dhatri Shankari T M , Sharanya Patil , Suresh Sundram This is my paper · ORCID
classification eess.IV
keywords underwaterimageenhancementbackscatterestimationmonoculardepthedge-preservinglossSobelfilterLaplacianofGaussianUIQMGPMAE
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

OceanLens is a deep-learning method for restoring color and detail in underwater photographs. The paper argues that by splitting the image into two physically meaningful parts—backscattered light and depth-dependent attenuation—and estimating each with its own convolutional network, enhancement can beat two published approaches (SeeThru and DeepSeeColor) without requiring multi-view reconstruction. The network is trained with an adaptive Huber backscatter loss and two edge-preserving losses (Sobel and Laplacian of Gaussian) that protect fine structure. The paper reports an average 65% reduction in Grayscale Patch Mean Angular Error and a 60% increase in the Underwater Image Quality Metric, and shows that adding convolution layers improves SSIM on the UIEB reference dataset. The broader promise is that a single image plus an off-the-shelf monocular depth estimate is enough to drive the correction.

What carries the argument

The load-bearing mechanism is the split of the underwater image into backscatter and attenuated direct signal, following the formation model $I^c = I_D^c + I_B^c$. A backscatter network turns the range/depth map into $\hat{I}_B$ via convolutional terms with CEAF/EAF activations; a deattenuation network turns the same map into $\hat{\alpha}_D^c(z)$, a sum of exponentials. The enhanced image is the product of the direct estimate and inverse attenuation. The objective that stabilizes this decomposition is a composite loss: an adaptive Huber backscatter term, saturation/intensity/variance terms, and Sobel plus Laplacian-of-Gaussian edge terms, which bias the network toward preserving luminance variance and high-frequency detail.

What would settle it

Take a SeeThru image and run OceanLens with its original SfM depth map, with MonoDepth2's predicted map, with Depth-Anything-V2-Large's predicted map, and with a deliberately scrambled depth map; if GPMAE and UIQM change little across all four conditions, the claim that depth drives the enhancement fails, whereas large degradation under scrambling would support it.

Watch

Extended reading notes

Core claim

On its own terms, the paper's discovery is that a physics-grounded network pair—one estimating backscatter $\hat{I}_B$ from a range map, the other estimating inverse attenuation $\hat{\alpha}_D^c(z)$—can reconstruct a direct image $\hat{I}_J = (I - \hat{I}_B)\,\hat{\alpha}_D^c(z)$ with better color and structure than the SeeThru and DeepSeeColor baselines. The backscatter estimate follows the exponential form $I_B = I_{B\infty}(1-\exp(-b_1 z)) + I'_B \exp(-b_2 z)$, implemented with complementary exponential and exponential activations; the attenuation estimate is a neural sum of exponentials. Three additions carry the reported gains: the adaptive Huber loss on backscatter, the Sobel and LoG edge losses in the deattenuation objective, and extra convolutional layers for detail. The paper also claims that depth maps from terrestrial-trained monocular models—MonoDepth2 and Depth-Anything-V2-Large—deliver performance in line with the original structure-from-motion range maps, with the larger Depth-Anything-V2-Large model giving the lowest angular errors on several SeeThru images.

Load-bearing premise

The method's results depend on pre-trained monocular depth models, trained on ordinary land images, producing depth maps that are accurate enough for underwater backscatter estimation, and the paper does not validate those predicted depths against ground-truth underwater range maps.

Editorial extensions

If this is right

  • OceanLens reports an average 65% lower Grayscale Patch Mean Angular Error and a 60% higher Underwater Image Quality Metric than the SeeThru and DeepSeeColor baselines on the SeeThru images.
  • Depth maps from MonoDepth2 and Depth-Anything-V2-Large can replace structure-from-motion range maps for the backscatter and attenuation estimates, with Depth-Anything-V2-Large giving the lowest angular errors on several images.
  • Adding convolutional layers improves structural fidelity: SSIM on the UIEB reference images rises, with Image3 up 6.3% and Image2 moving from negative to 0.0308, described as up to 12-15% overall SSIM improvement.
  • The Sobel and LoG edge losses measurably improve UIQM in the ablations, and the method runs fast enough for near real-time use at roughly 4-5 milliseconds per 7-12 MB image.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If terrestrial-trained monocular depth estimates truly suffice, then any single underwater image without range data becomes a candidate for physics-based enhancement; this could be tested by comparing OceanLens output using predicted depth against output using the SeeThru ground-truth range map on the same images.
  • The depth-source sensitivity visible in the reported tables suggests a natural next benchmark: matching the depth estimator to water conditions and lighting, since Depth-Anything-V2-Large wins on some images while the original depth map wins on others.
  • The backscatter-plus-attenuation split is not unique to water, so the same network structure with edge losses could be transferred to fog, haze, or turbid-media restoration by retraining the coefficient ranges.
  • The Sobel and LoG edge-loss combination could serve as a generic structural regularizer for image restoration, because it penalizes gradient mismatch and second-derivative mismatch separately.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 6 minor

Summary. The manuscript proposes OceanLens, a two-branch neural network for underwater image enhancement that estimates backscatter and deattenuation from an image and a depth map, using an adaptive Huber backscatter loss and a composite deattenuation loss with Sobel and LoG edge terms. The method is tested on SeeThru (5 images), US Virgin Islands (qualitative only), and UIEB (3 images), with claims of 65% GPMAE reduction, 60% UIQM gain, and 12-15% SSIM improvement over SeeThru and DeepSeeColor.

Significance. If the claims held, the work would provide a useful lightweight enhancement method and evidence that terrestrial monocular depth estimators transfer to underwater scenes. The authors also make code available. However, the reported aggregate numbers are not supported by the tables, the main quality metric is partially aligned with the training objective, and the evaluation set is too small to justify the claims.

major comments (4)
  1. [Abstract and Section IV-B] The abstract and conclusion state a 65% GPMAE reduction, 60% UIQM increase, and 12-15% SSIM improvement, but these aggregates are not derivable from Tables I-V. Table V reports only three UIEB images, with relative SSIM changes of +865% (Image1), a sign-crossing increase (Image2 from -0.0062 to 0.0308), and +6.3% (Image3); no entry is 12-15%. Tables I and III contain repeated per-patch entries for D5 only, and D5 shows GPMAE values exceeding ST/DSC and UIQM values below raw, yet the averaging rule for the 65% and 60% figures is not stated. The authors need to report per-image results, define the aggregation, and include variability estimates across training runs.
  2. [III-C.1, Eqs. (14)-(15) and IV-A.1, Eq. (21)] There is a circularity concern: Lsat with Isat_tar=1 and Lint with Itar push the output toward white/gray, while the GPMAE metric of Eq. (21) measures angular error against the (1,1,1) white vector. Minimizing these losses therefore directly reduces the reported evaluation metric by construction, so GPMAE improvements do not independently demonstrate color fidelity. The authors should either use a reference-based metric (e.g., on UIEB) for the SeeThru comparison or train without the saturation/intensity targets and show GPMAE still improves.
  3. [Section I and IV-B] The claim that pre-trained terrestrial monocular depth models (MonoDepth2, Depth-Anything-V2-Large) are suitable for underwater backscatter estimation is not validated. No quantitative comparison of MD/DA depth maps against the SeeThru ground-truth depth is provided, and Tables I-IV show large performance variations across depth sources (e.g., DA GPMAE 0.57 on D3 but 20-28 on D5 in Table III). A depth accuracy evaluation and its correlation with enhancement performance are needed before this assumption can support the method.
  4. [Tables I-V] The evaluation uses only five SeeThru images and three UIEB images, with no error bars, multiple runs, or statistical tests. The D5 failure is acknowledged in Section IV-B but is excluded from the aggregate claims without a stated rule. This sample size is insufficient to support the general superiority claims in the abstract; per-image results with variance and a pre-registered aggregation rule are required.
minor comments (6)
  1. [Section III-C] The word 'deattnuation' in the section heading should be 'deattenuation'.
  2. [Section IV-A] The heading 'Evalution Metrics' contains a typo; it should be 'Evaluation Metrics'.
  3. [Tables I and III] The D5 rows contain slash-separated values (e.g., 17/16/15/17) without explanation; clarify whether these are repeated patch measurements, different runs, or something else.
  4. [Fig. 7 caption] The caption contains 'OceanLeans', which should be 'OceanLens'.
  5. [Section IV-B] The UIEB results are reported for only three images with no image identifiers; provide the image names or indices so that the results are reproducible.
  6. [Sections III-B and III-C] Training hyperparameters (learning rate, optimizer, number of epochs, P, T, and the chosen values of delta and beta) are not specified; include them to allow replication.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the derivation chain is self-contained; the abstract's aggregate performance numbers are unsupported, but that is a correctness/support issue, not circularity.

full rationale

The paper's claimed contributions are (i) a physics-inspired two-network model with backscatter and deattenuation branches, (ii) novel loss terms including Sobel and LoG edge losses, (iii) use of pre-trained monocular depth maps, and (iv) empirical gains over Sea-Thru and DeepSeeColor. None of these derive a target from an input that already contains the target. Backscatter is estimated from the depth map via Eq. (6)-(7), and the direct image is reconstructed via Eq. (12); the network is trained with losses in Eq. (13)-(19). The evaluation metric GPMAE (Eq. 21) is never used as a training loss. It is true that Lsat and Lint push outputs toward white/gray, which aligns with GPMAE's definition of color fidelity (angle to the (1,1,1) vector), but this is an explicit loss-design choice, not a hidden re-importation of the evaluation metric. The comparisons against Sea-Thru and DeepSeeColor are external baselines, and the depth-map claim is an empirical assertion. There are no load-bearing self-citations; the architecture is credited to external prior work [7], [9]. The abstract's 65% GPMAE reduction, 60% UIQM increase, and 12-15% SSIM improvement are not recomputable from Tables I-V, and some table entries contradict the text (e.g., D5 failures, UIEB SSIM changes). However, this is a matter of evidential support and internal consistency, not circularity. No step in the paper reduces, by construction or by self-citation, to its own inputs. Therefore the correct circularity finding is no significant circularity (score 0).

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

The method relies on the standard underwater image formation model from Akkaynak and Treibitz, on activation functions borrowed from DeepSeeColor, and on a handful of hand-set loss targets and hyperparameters (delta, beta, intensity/saturation targets, layer counts). No new physical entities are introduced. The deepest issue is that several hyperparameters are chosen on the same test images that are later scored, which means part of the reported improvement is a selection effect.

free parameters (5)
  • Huber threshold delta = 0.5
    Chosen in Fig 7(c) as the value yielding superior UIQM on the SeeThru test images; this is hyperparameter tuning on the evaluation set.
  • Huber balance beta = not reported
    Hyperparameter in Eq (9) balancing quadratic and linear loss; no value or tuning procedure is given in the paper.
  • Intensity target Itar = e.g., 0.5 or 1.0 (not specified)
    Manual target in Eq (15) that pushes mean intensity toward mid-gray or white; affects all outputs and the GPMAE evaluation.
  • Saturation target Isat_tar = 1
    Manual target in Eq (14) that pushes pixel values toward [0,1] and toward 1; this directly reduces angular error on gray patches.
  • Number of convolutional layers (P/T) = N=2 for ODM, N=3 for DA, N=4 for MD (Fig 7a)
    Layer count is selected per depth map based on UIQM on the test set; the ablation chooses different N for different inputs.
assumptions (5)
  • domain assumption Underwater image formation model I_c = I_D + I_B with exponential attenuation and backscatter (Eqs 1-3)
    Taken from Akkaynak and Treibitz [4]; the entire network parameterization relies on this model.
  • domain assumption CEAF/EAF activations correctly implement the exponential model terms
    Adopted from DeepSeeColor [9]; Eqs (7)-(8) and (11) assume these activations parameterize the physics.
  • domain assumption Pre-trained terrestrial monocular depth models produce usable underwater depth maps without fine-tuning
    The depth input is obtained from MonoDepth2 and Depth-Anything-V2-Large, trained on in-air images; no underwater validation of depth accuracy is provided.
  • domain assumption No-reference metrics GPMAE and UIQM are valid proxies for underwater image quality
    The evaluation on SeeThru uses these metrics as ground truth for color and quality; the paper does not validate them against human perception.
  • standard math Standard neural network training converges to a good solution with the given losses
    The paper relies on standard SGD and backpropagation, not formally stated; this is a background assumption for any deep learning paper.

how reviews work

0 comments
Cite this review

Pith. "Pith review of OceanLens: An Adaptive Backscatter and Edge Correction using Deep Learning Model for Enhanced Underwater Imaging." pith.science (2026). https://pith.science/paper/KT3OXIVF

@misc{pith2026241113230,
  author       = {Pith},
  title        = {Pith review of: OceanLens: An Adaptive Backscatter and Edge Correction using Deep Learning Model for Enhanced Underwater Imaging},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KT3OXIVF}},
  note         = {Machine review of arXiv:2411.13230}
}
read the original abstract

Underwater environments pose significant challenges due to the selective absorption and scattering of light by water, which affects image clarity, contrast, and color fidelity. To overcome these, we introduce OceanLens, a method that models underwater image physics-encompassing both backscatter and attenuation-using neural networks. Our model incorporates adaptive backscatter and edge correction losses, specifically Sobel and LoG losses, to manage image variance and luminance, resulting in clearer and more accurate outputs. Additionally, we demonstrate the relevance of pre-trained monocular depth estimation models for generating underwater depth maps. Our evaluation compares the performance of various loss functions against state-of-the-art methods using the SeeThru dataset, revealing significant improvements. Specifically, we observe an average of 65% reduction in Grayscale Patch Mean Angular Error (GPMAE) and a 60% increase in the Underwater Image Quality Metric (UIQM) compared to the SeeThru and DeepSeeColor methods. Further, the results were improved with additional convolution layers that capture subtle image details more effectively with OceanLens. This architecture is validated on the UIEB dataset, with model performance assessed using Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index Measure (SSIM) metrics. OceanLens with multiple convolutional layers achieves up to 12-15% improvement in the SSIM.

Figures

Figures reproduced from arXiv: 2411.13230 by the authors.

Figure 1
Figure 1. Illustration of the OceanLens Architecture. The proposed OceanLens architecture operates with two key inputs: an underwater image and its corresponding range or depth map. These inputs are processed through two specialized neural networks—the backscatter correction [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Network Architecture for Backscatter: The network takes Range map Z as input and predicts backscatter image IˆB, with the kernel parameters in each convolution layer corresponding to those of the backscatter estimation model in (6). where the second term models a small residual component of the direct signal. Here, I c B∞ and I c ′ B are network parameters. The BackscatterNet network is designed to estimate the dire… view at source ↗
Figure 3
Figure 3. Network Architecture for Deattenuation αˆ c D(Z): The network generates the αˆ c D(Z) from range map Z, with the kernel parameters a c pi in each convolution layer corresponding to the parameters of the attenuation coefficient a c D function as described in (10). C. Deattenuation Modeling The deattenuation model aims to correct the observed image based on depth and attenuation factors. The coefficient of attenuation… view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: (a) Raw Images from SeeThru [4] (Row 1 - 4) and US Virgin Islands [37] dataset (Row 5). (b), (c), (d) Enhanced images obtained through OceanLens with one convolutional layer using ODM, MD and DA respectively. 85.5%, 81.3%, and 91.42% reduction as compared to DSC as see…
Figure 5
Figure 5. Figure 5: (a) Raw Images from SeeThru (Row 1-4) and US Virgin Islands dataset (Row 5). (b), (c), (d) Enhanced images obtained through OceanLens with multiple convolutional layer using ODM, MD and DA respectively. (a) (b) (c) [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: (a) Raw Images from UIEB dataset [26]. (b) Corresponding reference images. (c) Enhanced images - OceanLens with multiple convolutional layers using DA depth map. while also indicating areas for further refinement. C. Ablation Study The results presented in the graphs i…
Figure 7
Figure 7. Figure 7: Variation of UIQM scores with multiple convolutional layers in OceanLeans: (a) With different depth maps, (b) With and without the use of Sobel and LoG loss functions and (c) Variation of UIQM scores with different δ, an Adaptive Huber loss function parameter. [3] Y. W…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

40 extracted references · 32 canonical work pages

  1. [1]

    Algorithms for improving the quality of underwater optical images: A comprehensive review,

    X. Shuang, J. Zhang, and Y . Tian, “Algorithms for improving the quality of underwater optical images: A comprehensive review,”Signal Processing, p. 109408, 2024

    work page 2024

  2. [2]

    What is the space of attenuation coefficients in underwater computer vision?,

    D. Akkaynak, T. Treibitz, T. Shlesinger, Y . Loya, R. Tamir, and D. Iluz, “What is the space of attenuation coefficients in underwater computer vision?,” in Proceedings of the IEEE conference on computer vision and pattern recognition , pp. 4931–4940, 2017. (a) (b) (c) Fig. 7: Variation of UIQM scores with multiple convolutional layers in OceanLeans: (a) ...

    work page 2017

  3. [3]

    An experimental-based review of image enhancement and image restoration methods for underwater imaging,

    Y . Wang, W. Song, G. Fortino, L. Qi, W. Zhang, and A. Liotta, “An experimental-based review of image enhancement and image restoration methods for underwater imaging,” IEEE Access , vol. 7, pp. 140233–140251, 2019

    work page 2019

  4. [4]

    A revised underwater image formation model,

    D. Akkaynak and T. Treibitz, “A revised underwater image formation model,” in Proceedings of the IEEE conference on computer vision and pattern recognition , pp. 6723–6732, 2018

    work page 2018

  5. [5]

    Active reward learning for co- robotic vision based exploration in bandwidth limited environments,

    S. Jamieson, J. P. How, and Y . Girdhar, “Active reward learning for co- robotic vision based exploration in bandwidth limited environments,” in 2020 IEEE International Conference on Robotics and Automation (ICRA), pp. 1806–1812, IEEE, 2020

    work page 2020

  6. [6]

    Underwater image quality improvement via color, detail, and contrast restoration,

    Z. Liang, W. Zhang, R. Ruan, P. Zhuang, X. Xie, and C. Li, “Underwater image quality improvement via color, detail, and contrast restoration,” IEEE Transactions on Circuits and Systems for Video Technology, 2023

    work page 2023

  7. [7]

    Sea-thru: A method for removing water from underwater images,

    D. Akkaynak and T. Treibitz, “Sea-thru: A method for removing water from underwater images,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pp. 1682–1691, 2019

    work page 2019

  8. [8]

    Semi-supervised advancement of underwater visual quality,

    H. Zhu, X. Han, and Y . Tao, “Semi-supervised advancement of underwater visual quality,” Measurement Science and Technology , vol. 32, no. 1, p. 015404, 2020

    work page 2020

Show all 40 references
  1. [9]

    Deepseecolor: Realtime adaptive color correction for autonomous underwater vehicles via deep learning methods,

    S. Jamieson, J. P. How, and Y . Girdhar, “Deepseecolor: Realtime adaptive color correction for autonomous underwater vehicles via deep learning methods,” in 2023 IEEE International Conference on Robotics and Automation (ICRA) , pp. 3095–3101, 2023

    2023

  2. [10]

    Digging into self-supervised monocular depth estimation,

    C. Godard, O. Mac Aodha, M. Firman, and G. J. Brostow, “Digging into self-supervised monocular depth estimation,” in Proceedings of the IEEE/CVF international conference on computer vision, pp. 3828– 3838, 2019

    2019

  3. [11]

    Depth anything v2,

    L. Yang, B. Kang, Z. Huang, Z. Zhao, X. Xu, J. Feng, and H. Zhao, “Depth anything v2,” arXiv preprint arXiv:2406.09414 , 2024

    2024 arXiv

  4. [12]

    Diving into haze-lines: Color restoration of underwater images,

    D. Berman, T. Treibitz, and S. Avidan, “Diving into haze-lines: Color restoration of underwater images,” in Proc. British Machine Vision Conference (BMVC), vol. 1, p. 2, 2017

    2017

  5. [13]

    A review on intelligence dehazing and color restoration for underwater images,

    M. Han, Z. Lyu, T. Qiu, and M. Xu, “A review on intelligence dehazing and color restoration for underwater images,” IEEE Transactions on Systems, Man, and Cybernetics: Systems , vol. 50, no. 5, pp. 1820– 1832, 2018

    2018

  6. [14]

    Underwater image enhancement using white balance, usm and clhe,

    K. Sanila, A. A. Balakrishnan, and M. Supriya, “Underwater image enhancement using white balance, usm and clhe,” in 2019 Interna- tional Symposium on Ocean Technology (SYMPOL) , pp. 106–116, IEEE, 2019

    2019

  7. [15]

    A single image dehazing method based on adaptive gamma correction,

    N. M. Hong and N. Chi Thanh, “A single image dehazing method based on adaptive gamma correction,” in 2019 6th NAFOSTED Con- ference on Information and Computer Science (NICS) , pp. 558–562, 2019

    2019

  8. [16]

    Mixture contrast limited adaptive histogram equalization for underwater image enhancement,

    M. S. Hitam, E. A. Awalludin, W. N. J. H. W. Yussof, and Z. Ba- chok, “Mixture contrast limited adaptive histogram equalization for underwater image enhancement,” in 2013 International conference on computer applications technology (ICCAT) , pp. 1–5, IEEE, 2013

    2013

  9. [17]

    Single image haze removal using dark channel prior,

    K. He, J. Sun, and X. Tang, “Single image haze removal using dark channel prior,” IEEE transactions on pattern analysis and machine intelligence, vol. 33, no. 12, pp. 2341–2353, 2010

    2010

  10. [18]

    Underwater image restoration based on improved background light estimation and automatic white balance,

    C. Li and X. Zhang, “Underwater image restoration based on improved background light estimation and automatic white balance,” in 2018 11th International Congress on Image and Signal Processing, BioMed- ical Engineering and Informatics (CISP-BMEI) , pp. 1–5, IEEE, 2018

    2018

  11. [19]

    Color map optimization for 3d reconstruc- tion with consumer depth cameras,

    Q.-Y . Zhou and V . Koltun, “Color map optimization for 3d reconstruc- tion with consumer depth cameras,” ACM Transactions on Graphics (ToG), vol. 33, no. 4, pp. 1–10, 2014

    2014

  12. [20]

    Impact of colour on robustness of deep neural networks,

    K. De and M. Pedersen, “Impact of colour on robustness of deep neural networks,” in Proceedings of the IEEE/CVF international conference on computer vision , pp. 21–30, 2021

    2021

  13. [21]

    Underwater image enhancement method via multi-feature prior fusion,

    J. Zhou, D. Zhang, and W. Zhang, “Underwater image enhancement method via multi-feature prior fusion,” Applied Intelligence , vol. 52, no. 14, pp. 16435–16457, 2022

    2022

  14. [22]

    Hfm: A hy- brid fusion method for underwater image enhancement,

    S. An, L. Xu, I. Senior Member, Z. Deng, and H. Zhang, “Hfm: A hy- brid fusion method for underwater image enhancement,” Engineering Applications of Artificial Intelligence , vol. 127, p. 107219, 2024

    2024

  15. [23]

    Underwater image enhancement based on histogram-equalization approximation using physics-based dichromatic modeling,

    Y .-T. Peng, Y .-R. Chen, Z. Chen, J.-H. Wang, and S.-C. Huang, “Underwater image enhancement based on histogram-equalization approximation using physics-based dichromatic modeling,” Sensors, vol. 22, no. 6, p. 2168, 2022

    2022

  16. [24]

    Phish-net: Physics inspired system for high resolution underwater image enhancement,

    A. Chandrasekar, M. Sreenivas, and S. Biswas, “Phish-net: Physics inspired system for high resolution underwater image enhancement,” in Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision , pp. 1506–1516, 2024

    2024

  17. [25]

    Deep unfolding network with physics-based priors for underwater image enhancement,

    T. T. Pham, T. T. N. Mai, and C. Lee, “Deep unfolding network with physics-based priors for underwater image enhancement,” in2023 IEEE International Conference on Image Processing (ICIP) , pp. 46– 50, 2023

    2023

  18. [26]

    An underwater image enhancement benchmark dataset and beyond,

    C. Li, C. Guo, W. Ren, R. Cong, J. Hou, S. Kwong, and D. Tao, “An underwater image enhancement benchmark dataset and beyond,” IEEE transactions on image processing , vol. 29, pp. 4376–4389, 2019

    2019

  19. [27]

    Fast underwater image enhancement for improved visual perception,

    M. J. Islam, Y . Xia, and J. Sattar, “Fast underwater image enhancement for improved visual perception,” IEEE Robotics and Automation Letters, vol. 5, no. 2, pp. 3227–3234, 2020

    2020

  20. [28]

    Watergan: Unsupervised generative network to enable real-time color correction of monocular underwater images,

    J. Li, K. A. Skinner, R. M. Eustice, and M. Johnson-Roberson, “Watergan: Unsupervised generative network to enable real-time color correction of monocular underwater images,” IEEE Robotics and Automation letters, vol. 3, no. 1, pp. 387–394, 2017

    2017

  21. [29]

    Udepth: Fast monocular depth estimation for visually-guided underwater robots,

    B. Yu, J. Wu, and M. J. Islam, “Udepth: Fast monocular depth estimation for visually-guided underwater robots,” in 2023 IEEE Inter- national Conference on Robotics and Automation (ICRA) , pp. 3116– 3123, IEEE, 2023

    2023

  22. [30]

    Semantic attention and relative scene depth-guided network for underwater image enhancement,

    T. Chen, N. Wang, Y . Chen, X. Kong, Y . Lin, H. Zhao, and H. R. Karimi, “Semantic attention and relative scene depth-guided network for underwater image enhancement,” Engineering Applications of Artificial Intelligence, vol. 123, p. 106532, 2023

    2023

  23. [31]

    Underwater sin- gle image color restoration using haze-lines and a new quantitative dataset,

    D. Berman, D. Levy, S. Avidan, and T. Treibitz, “Underwater sin- gle image color restoration using haze-lines and a new quantitative dataset,” IEEE transactions on pattern analysis and machine intelli- gence, vol. 43, no. 8, pp. 2822–2837, 2020

    2020

  24. [32]

    True color correction of autonomous underwater vehicle imagery,

    M. Bryson, M. Johnson-Roberson, O. Pizarro, and S. B. Williams, “True color correction of autonomous underwater vehicle imagery,” Journal of Field Robotics , vol. 33, no. 6, pp. 853–874, 2016

    2016

  25. [33]

    Overview of underwater 3d reconstruction technology based on optical images,

    K. Hu, T. Wang, C. Shen, C. Weng, F. Zhou, M. Xia, and L. Weng, “Overview of underwater 3d reconstruction technology based on optical images,” Journal of Marine Science and Engineering , vol. 11, no. 5, p. 949, 2023

    2023

  26. [34]

    Underwater image restora- tion via background light estimation and depth map optimization,

    D. Liu, J. Zhou, X. Xie, Z. Lin, and Y . Lin, “Underwater image restora- tion via background light estimation and depth map optimization,” Optics Express, vol. 30, no. 16, pp. 29099–29116, 2022

    2022

  27. [35]

    Underwater image restoration via depth map and illumination estimation based on a single image,

    J. Zhou, T. Yang, W. Ren, D. Zhang, and W. Zhang, “Underwater image restoration via depth map and illumination estimation based on a single image,” Optics Express , vol. 29, no. 19, pp. 29864–29886, 2021

    2021

  28. [36]

    Adaptive huber regression,

    Q. Sun, W.-X. Zhou, and J. Fan, “Adaptive huber regression,” Journal of the American Statistical Association , vol. 115, no. 529, pp. 254– 265, 2020

    2020

  29. [37]

    Curee: A curious un- derwater robot for ecosystem exploration,

    Y . Girdhar, N. McGuire, L. Cai, S. Jamieson, S. McCammon, B. Claus, J. E. San Soucie, J. E. Todd, and T. A. Mooney, “Curee: A curious un- derwater robot for ecosystem exploration,” in 2023 IEEE International Conference on Robotics and Automation (ICRA) , pp. 11411–11417, IEEE, 2023

    2023

  30. [38]

    Depth anything: Unleashing the power of large-scale unlabeled data,

    L. Yang, B. Kang, Z. Huang, X. Xu, J. Feng, and H. Zhao, “Depth anything: Unleashing the power of large-scale unlabeled data,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 10371–10381, 2024

    2024

  31. [39]

    An underwater color image quality evaluation metric,

    M. Yang and A. Sowmya, “An underwater color image quality evaluation metric,” IEEE Transactions on Image Processing , vol. 24, no. 12, pp. 6062–6071, 2015

    2015

  32. [40]

    Human-visual-system-inspired underwater image quality measures,

    K. Panetta, C. Gao, and S. Agaian, “Human-visual-system-inspired underwater image quality measures,” IEEE Journal of Oceanic Engi- neering, vol. 41, no. 3, pp. 541–551, 2015

    2015

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.