Pith. sign in

REVIEW 2 major objections 4 minor 135 references

Reversing Flow for Image Restoration

T0 review · 2 major / 4 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read Image restoration can be done by reversing a learned deterministic degradation flow, cutting sampling to four steps or fewer.

desk verdict Strong empirical flow-matching restoration with four-step sampling, but the auxiliary-variable and entropy-preservation claims do not survive the paper's own equations. read the letter →

arxiv 2506.16961 v1 pith:D7UN6VMF submitted 2025-06-20 cs.CV eess.IV

Haina Qin , Wenyang Luo , Libin Wang , Dandan Zheng , Jingdong Chen , Ming Yang , Bing Li , Weiming Hu This is my paper · ORCID
classification cs.CVeess.IV
keywords imagerestorationcontinuousnormalizingflowmatchingauxiliaryvariabledegradationmodelingfew-stepinferenceill-posedinverseproblems
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

This paper proposes ResFlow, a framework that treats image degradation—from clean high-quality image to corrupted low-quality image—as a deterministic path rather than a stochastic diffusion process. Because ordinary deterministic flows preserve information while real degradation destroys it, ResFlow augments the state with an auxiliary variable that is meant to encode what is lost, making the path invertible. The augmented flow is learned by matching the velocity field of a simple linear interpolation between HQ and LQ images, and restoration is done by solving the learned ODE backward from the degraded image in fewer than four steps. If the central claim holds, image restoration becomes much cheaper and faster than diffusion-based methods while matching or exceeding their quality across tasks like denoising, dehazing, deraining, and deblurring.

What carries the argument

The central object is the augmented continuous normalizing flow on the concatenated state $z_t = [x_t; y_t]$, governed by the ODE $\partial z_t/\partial t = v_\theta(x_t,y_t,t)$. The data component $x_t$ interpolates linearly between HQ and LQ, while the auxiliary component $y_t$ evolves with the schedule $\sigma^y_t = \beta(1-t+\beta)^{-1}$; together they define the ground-truth velocity $\dot{z}_t$ used in the flow-matching loss. The auxiliary variable is the mechanism that carries the argument: conceptually it encodes the information lost in degradation, allowing the learned velocity network to pick one HQ image among the many consistent with a given LQ image. Training never simulates the ODE; instead the velocity field is regressed pointwise against the analytic derivative of the interpolation, and inference integrates the learned field backward with Euler steps.

What would settle it

For a fixed degraded image, sample two auxiliary vectors $y_1 \sim N(0,I)$ and run the backward ODE; if the two restored outputs do not differ in the ambiguous regions, or if one is clearly hallucinated, the disambiguation claim fails. Alternatively, compute the differential entropy of $z_t$ along the schedule (8); if it changes with $t$, the entropy-preserving property is violated.

Watch

Extended reading notes

Core claim

ResFlow's central claim is that image restoration can be cast as the reverse of a deterministic continuous normalizing flow that is augmented with an auxiliary random variable. The state space is $z_t = [x_t; y_t]$, where $x_t$ moves from the HQ image $x_0$ to the LQ image $x_1$ by linear interpolation $x_t = (1-t)x_0 + t x_1$, and $y_t$ follows the schedule $\sigma^y_t = \beta/(1-t+\beta)$ with $y_0 = 0$ and $y_1$ a standard Gaussian. The velocity field of this augmented flow is learned by minimizing the flow-matching loss $\mathbb{E}\int_0^1 \lambda(t) \|v_\theta(x_t,y_t,t) - \dot{z}_t\|^2 dt$, with a loss weight that emphasizes times near $t=1$. At inference, the corrupted image $x_1$ is taken as the starting point, an auxiliary sample $y_1\sim N(0,I)$ is drawn, and the ODE is integrated backward to $t=0$; the auxiliary variable is what the paper argues disambiguates which of several plausible clean images produced the observed degradation, and the schedule is described as entropy-preserving so that the reversal is a well-defined reversible process. The paper reports state-of-the-art PSNR/SSIM/LPIPS on several benchmarks with four or fewer sampling steps.

Load-bearing premise

The claim rests on the assumption that the learned velocity network, trained to match the linear-interpolation velocities, produces a deterministic coupling between the random auxiliary variable and the clean image that genuinely disambiguates the ill-posed map—so that every sampled auxiliary input yields a valid, sharp restoration rather than an inconsistent or averaged one.

Editorial extensions

If this is right

  • Restoration can be completed in fewer than four sampling steps, making inference far cheaper than diffusion models that typically need dozens or hundreds of steps.
  • The same trained velocity field and inference scheme apply across degradation types (snow, rain, haze, noise, blur, JPEG artifacts) without task-specific modifications, since the degradation path is defined by the data pair rather than a hand-designed operator.
  • The auxiliary variable at $t=1$ can be sampled as a Gaussian, so the method inherently offers a way to produce multiple plausible restored images for one degraded input.
  • Because training only matches the velocity of a known interpolation, it avoids simulating the ODE and therefore has low per-iteration cost, which the paper argues makes it practical for real-world deployment.
  • The loss weighting that emphasizes $t$ near 1 is what the paper credits for preserving sharp details that single-step methods average away, so one-step baselines should be dominated on strong degradations.

Reading between the lines

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

  • If the deterministic coupling between $y_1$ and $x_0$ is as clean as claimed, the same framework could be adapted for conditional image generation, where the auxiliary variable plays the role of a latent code that selects among plausible restorations; a direct test would be comparing the diversity of outputs on a known one-to-many degradation model.
  • The paper's own derivation in the supplementary material shows that exact entropy preservation would require $\sigma^y_t = \beta/(1-t)$, which is singular at $t=1$; the implemented schedule $\beta/(1-t+\beta)$ is a regularized approximation, so the entropy-preserving property should be understood approximately, and its empirical role can be tested by swapping in the singular schedule with a small
  • The linear interpolation for $x_t$ ties ResFlow to rectified-flow-style straight paths, which suggests that the four-step budget could be reduced to one step via distillation, a direction the paper does not explore.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 4 minor

Summary. The paper proposes ResFlow, a flow-matching framework for image restoration that models the degradation from high-quality (HQ) to low-quality (LQ) images as a deterministic continuous normalizing flow. To handle the ill-posedness of restoration, the state is augmented with an auxiliary process y_t, claimed to disambiguate the velocity field. The method uses a linear interpolation schedule for x_t and a proposed "entropy-preserving" schedule for y_t, trains a velocity network by matching the conditional flow-matching objective (Eq. 9), and restores images by solving the ODE backward from t=1 to t=0 with a sampled y_1. Experiments on desnowing, deraining, dehazing, denoising, defocus deblurring, and JPEG artifact removal report state-of-the-art results with fewer than four sampling steps.

Significance. The empirical contribution is substantial: ResFlow consistently outperforms strong baselines across multiple restoration tasks and datasets, with up to 2.24 dB PSNR gains over Restormer on SIDD and strong results with only four sampling steps. The paper also provides thorough ablations, implementation details, and supplementary visualizations, which support reproducibility. However, the theoretical centerpiece—the entropy-preserving schedule and the claim that the auxiliary variable disambiguates the uncertain inverse map—is contradicted by the paper's own equations. Given that these claims are central to the paper's novelty narrative, the significance of the contribution depends on whether they can be corrected or demoted in revision.

major comments (2)
  1. [Sec. 3.2, Eqs. (9)-(10)] The claim that the trained velocity network "induces a deterministic coupling" between y_1 and the HQ image x_0 is not supported by the training objective. In Eq. (9), y_1 is sampled as N(0,I) independently of (x_0,x_1), and y_0 = 0, so y_t = σ^y_t y_1 is independent of (x_0,x_1) for all t. The target x-velocity is \dot x_t = x_1 - x_0. Minimizing the flow-matching loss (9) for the x-component yields the conditional expectation E[x_1 - x_0 | x_t, y_t], which, by the independence of y_t, reduces to E[x_1 - x_0 | x_t]. Thus, at the optimum, the learned x-velocity is independent of y_t, and the ODE (6) starting from a fixed LQ image x_1 produces the same x_0 for every sampled y_1. The different outputs for different auxiliary variables shown in Fig. 11 are therefore an optimization artifact rather than a learned disambiguation of the ill-posed inverse. This directly undermines the central mechanism claimed in Sec. 3.2 and in the abstract.
  2. [Sec. 3.2, Eq. (8) and Supplementary Sec. 7, Eqs. (33)-(34)] The "entropy-preserving" degradation schedule is internally inconsistent. The derivation in the supplementary sets H(z_t)=H(z_s), leading to Eq. (33), whose solution is σ^y_t = β/(1-t), singular at t=1. The schedule actually used, σ^y_t = β/(1-t+β), does not satisfy Eq. (33): with β=10, ln(1-t)+ln σ^y_t equals ln(10/11) at t=0 but ln(5/10.5) at t=0.5, so the entropy of z_t is not constant. In addition, σ^y_0 = β/(1+β) ≠ 0, violating the boundary condition y_0 = 0 stated in Eq. (3) and the endpoint requirement σ_0 = 0 in Eq. (5). The abstract and introduction claim that ResFlow "adopts entropy-preserving flow paths," but this property is contradicted by the paper's own equations. The schedule may still be a useful heuristic, but it cannot be justified by the constant-entropy derivation as written.
minor comments (4)
  1. [Sec. 4.1] The text lists "draining" as a task name; this should read "deraining."
  2. [Sec. 3.3] The sentence "Optimizing Eq. (10) is efficient" appears to reference the wrong equation; the optimization objective is Eq. (9), while Eq. (10) defines the target velocity components.
  3. [Supplementary, Fig. 11 caption] The caption of Fig. 11 is identical to that of Fig. 10 and describes defocus deblurring, but the surrounding text and figure content discuss the effect of auxiliary variables on restored outputs. The caption should be corrected.
  4. [Supplementary, Sec. 8] The description of the Snow100K dataset contains the word "nozomura," which appears to be a typographical error; also, "Outdoor-Rain [49] contains 8100 images for training and 900 images for testing;." has an extra semicolon.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the core velocity-matching objective is a supervised regression on external paired data, and no fitted constant or self-citation chain determines the reported restoration outputs.

full rationale

ResFlow's derivation chain is not circular. The forward path z_t is defined by explicit interpolation schedules (Eqs. 4, 7, 8), and the velocity field is learned by matching the closed-form derivatives of those interpolations (Eq. 9). The targets ẋ_t and ẏ_t are computed from the paired training images and the chosen schedule; they are not fitted parameters renamed as predictions. The reported restoration quality is benchmarked on held-out datasets, so the central performance claim has independent empirical content. The auxiliary variable y_t is sampled from a fixed Gaussian at inference, and the paper's claim that the trained network learns a deterministic coupling between y_1 and x_0 is an empirical/architectural assertion, not a tautology: the objective does not define y_1 in terms of x_0, and the coupling is not enforced by construction. Whether that assertion actually holds is a correctness question, not a circularity one. Similarly, the entropy-preserving schedule is derived in the supplementary from a constant-entropy equation (Eq. 33), yielding σ_y^t = β/(1-t), which is singular at t=1; the paper then substitutes β/(1-t+β), and this regularized form does not satisfy the same constant-entropy equation. That is an internal-consistency issue, not a circular reduction: the schedule is a hand-selected hyperparameter, not an output of the model or a restatement of the result. The paper also contains no load-bearing self-citation chain; references to flow matching and rectified flow are external prior work, and no uniqueness theorem from the authors is invoked to force the design. The lack of statistical validation for the disambiguation effect (Fig. 11 shows only a few examples) is a rigor concern, but it does not make any derivation equivalent to its inputs. The natural finding is therefore no significant circularity.

Assumptions & free parameters 2 free parameters · 4 assumptions · 1 invented entities

The central mechanism is a standard flow-matching pipeline; the paper adds a hand-designed schedule and a learned latent. The entropy-preserving derivation is internally inconsistent, which weakens the theoretical grounding, but no fitted parameter directly forces the empirical results.

free parameters (2)
  • beta (beta=10) = 10
    Controls the auxiliary variable schedule sigma_y = beta / (1 - t + beta); chosen by hand without tuning, but the entropy-preserving derivation instead gives beta / (1 - t), so this value and the modified form are ad hoc.
  • gamma (gamma=1.75) = 1.75
    Exponent of the cosine loss weighting lambda(t); chosen empirically to emphasize timesteps near 1.
assumptions (4)
  • standard math Mutual information is invariant under invertible transformations (Lemma 1, from Kraskov et al.).
    Used in the proof of Proposition 1 to conclude that ODE flows preserve MI; accepted as a standard information-theoretic result.
  • domain assumption The degradation from HQ to LQ can be represented by a deterministic linear interpolation z_t = alpha_t z_0 + sigma_t z_1.
    Eq. (4)-(5); this is a modeling choice, not derived from physical degradation models, and it is the core of the flow-matching construction.
  • ad hoc to paper For the entropy derivation, HQ images are uniformly distributed on [0,1] and LQ images are Dirac at 0.
    Supplementary Sec. 7; these extreme assumptions are used only to derive the schedule and are not validated against real data.
  • domain assumption The trained velocity network induces a deterministic coupling between y and x_0 that disambiguates the uncertainty.
    Sec. 3.2 and Sec. 11; this is asserted and illustrated with examples, but not proven to hold in general.
invented entities (1)
  • Auxiliary variable y_t
    purpose: Encodes the information lost during degradation so the augmented ODE is reversible and the reverse process selects a specific HQ image.
    Figure 11 shows that different sampled y_1 lead to different restored details, providing an in-model handle, but there is no external or physical evidence for what y encodes.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Reversing Flow for Image Restoration." pith.science (2026). https://pith.science/paper/D7UN6VMF

@misc{pith2026250616961,
  author       = {Pith},
  title        = {Pith review of: Reversing Flow for Image Restoration},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/D7UN6VMF}},
  note         = {Machine review of arXiv:2506.16961}
}
read the original abstract

Image restoration aims to recover high-quality (HQ) images from degraded low-quality (LQ) ones by reversing the effects of degradation. Existing generative models for image restoration, including diffusion and score-based models, often treat the degradation process as a stochastic transformation, which introduces inefficiency and complexity. In this work, we propose ResFlow, a novel image restoration framework that models the degradation process as a deterministic path using continuous normalizing flows. ResFlow augments the degradation process with an auxiliary process that disambiguates the uncertainty in HQ prediction to enable reversible modeling of the degradation process. ResFlow adopts entropy-preserving flow paths and learns the augmented degradation flow by matching the velocity field. ResFlow significantly improves the performance and speed of image restoration, completing the task in fewer than four sampling steps. Extensive experiments demonstrate that ResFlow achieves state-of-the-art results across various image restoration benchmarks, offering a practical and efficient solution for real-world applications.

Figures

Figures reproduced from arXiv: 2506.16961 by the authors.

Figure 1
Figure 1. Image restoration is an ill-posed problem. The degra [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Framework of ResFlow. RV stands for random variable. The state zt consists of a data component xt that transits between HQ and LQ images and an auxiliary component yt that disambiguates the velocity to ensure invertibility. The forward process is defined by interpolation, while the reverse process is learned by matching the velocity field. The lower part depicts the transition of ResFlow. Image degradation is usuall… view at source ↗
Figure 3
Figure 3. Dehazing, Deraining, and Desnowing results. The part of the image is methodized to observe the local details clearly. From left [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (8 more)
Figure 4
Figure 4. Figure 4: Single-image defocus deblurring results on the DPDD [ [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Reserve process performance curves, averaged on 32 samples from the desnowing dataset. The right shows the intermediate [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: shows the curves of the degradation schedule σ y t un￾der various β values. We fix β = 10 for all experiments without tuning [PITH_FULL_IMAGE:figures/full_fig_p016_6.png]
Figure 7
Figure 7. Figure 7: Model architecture. All experiments adopt the same U-Net architecture from [31] as the backbone. The input to the U-Net is the xt start￾ing from x1 as the LQ image. The output of the U-Net is the velocity vθ(xt, yt, t). We remove the class-label condi￾tioning and condi…
Figure 8
Figure 8. Figure 8: Visual results of synthesized datasets. Input Output Reference Input Output Reference Haze Snow Rain Defocus Blur [PITH_FULL_IMAGE:figures/full_fig_p019_8.png]
Figure 9
Figure 9. Figure 9: Visual results of real-world datasets [PITH_FULL_IMAGE:figures/full_fig_p019_9.png]
Figure 10
Figure 10. Figure 10: Extra single-image defocus deblurring results on the DPDD [ [PITH_FULL_IMAGE:figures/full_fig_p020_10.png]
Figure 11
Figure 11. Figure 11: Extra single-image defocus deblurring results on the DPDD [ [PITH_FULL_IMAGE:figures/full_fig_p020_11.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

135 extracted references · 54 canonical work pages

  1. [1]

    A high-quality denoising dataset for smartphone cameras

    Abdelrahman Abdelhamed, Stephen Lin, and Michael S Brown. A high-quality denoising dataset for smartphone cameras. InProceedings of the IEEE conference on com- puter vision and pattern recognition, pages 1692–1700,

  2. [2]

    Defocus de- blurring using dual-pixel data

    Abdullah Abuolaim and Michael S Brown. Defocus de- blurring using dual-pixel data. InComputer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part X 16, pages 111–126. Springer, 2020. 5, 6, 7, 3, 4

    2020

  3. [3]

    Ntire 2017 challenge on single image super-resolution: Dataset and study

    Eirikur Agustsson and Radu Timofte. Ntire 2017 challenge on single image super-resolution: Dataset and study. InThe IEEE Conference on Computer Vision and Pattern Recog- nition (CVPR) Workshops, 2017. 3

    2017

  4. [4]

    Dense-haze: A benchmark for image dehazing with dense-haze and haze-free images

    Codruta O Ancuti, Cosmin Ancuti, Mateu Sbert, and Radu Timofte. Dense-haze: A benchmark for image dehazing with dense-haze and haze-free images. In2019 IEEE in- ternational conference on image processing (ICIP), pages 1014–1018. IEEE, 2019. 5, 7, 2, 4

    2019

  5. [5]

    Nh-haze: An image dehazing benchmark with non- homogeneous hazy and haze-free images

    Codruta O Ancuti, Cosmin Ancuti, and Radu Timo- fte. Nh-haze: An image dehazing benchmark with non- homogeneous hazy and haze-free images. InProceedings of the IEEE/CVF conference on computer vision and pat- tern recognition workshops, pages 444–445, 2020. 5, 6, 7, 2, 3, 4

    2020

  6. [6]

    Contour detection and hierarchical image segmentation.IEEE transactions on pattern analysis and machine intelligence, 33(5):898–916, 2010

    Pablo Arbelaez, Michael Maire, Charless Fowlkes, and Ji- tendra Malik. Contour detection and hierarchical image segmentation.IEEE transactions on pattern analysis and machine intelligence, 33(5):898–916, 2010. 3

    2010

  7. [7]

    Wasserstein generative adversarial networks

    Martin Arjovsky, Soumith Chintala, and L ´eon Bottou. Wasserstein generative adversarial networks. InInterna- tional conference on machine learning, pages 214–223. PMLR, 2017. 2

    2017

  8. [8]

    Self-guided image dehazing using progressive feature fu- sion.IEEE Transactions on Image Processing, 31:1217– 1229, 2022

    Haoran Bai, Jinshan Pan, Xinguang Xiang, and Jinhui Tang. Self-guided image dehazing using progressive feature fu- sion.IEEE Transactions on Image Processing, 31:1217– 1229, 2022. 5

    2022

Show all 135 references
  1. [9]

    Digital image restoration.IEEE signal processing magazine, 14(2):24– 41, 1997

    Mark R Banham and Aggelos K Katsaggelos. Digital image restoration.IEEE signal processing magazine, 14(2):24– 41, 1997. 2

    1997

  2. [10]

    Cold diffusion: Inverting arbitrary im- age transforms without noise.Advances in Neural Informa- tion Processing Systems, 36:41259–41282, 2023

    Arpit Bansal, Eitan Borgnia, Hong-Min Chu, Jie Li, Hamid Kazemi, Furong Huang, Micah Goldblum, Jonas Geiping, and Tom Goldstein. Cold diffusion: Inverting arbitrary im- age transforms without noise.Advances in Neural Informa- tion Processing Systems, 36:41259–41282, 2023. 3

    2023

  3. [11]

    An intuitive proof of the data processing inequality.arXiv preprint arXiv:1107.0740, 2011

    Normand J Beaudry and Renato Renner. An intuitive proof of the data processing inequality.arXiv preprint arXiv:1107.0740, 2011. 1, 4

    2011 arXiv

  4. [12]

    The perception-distortion tradeoff

    Yochai Blau and Tomer Michaeli. The perception-distortion tradeoff. InProceedings of the IEEE conference on com- puter vision and pattern recognition, pages 6228–6237,

  5. [13]

    Dehazenet: An end-to-end system for single image haze removal.IEEE transactions on image process- ing, 25(11):5187–5198, 2016

    Bolun Cai, Xiangmin Xu, Kui Jia, Chunmei Qing, and Dacheng Tao. Dehazenet: An end-to-end system for single image haze removal.IEEE transactions on image process- ing, 25(11):5187–5198, 2016. 5, 6

    2016

  6. [14]

    Simple baselines for image restoration

    Liangyu Chen, Xiaojie Chu, Xiangyu Zhang, and Jian Sun. Simple baselines for image restoration. InEuropean con- ference on computer vision, pages 17–33. Springer, 2022. 5, 6, 7

    2022

  7. [15]

    Neural ordinary differential equa- tions.Advances in neural information processing systems, 31, 2018

    Ricky TQ Chen, Yulia Rubanova, Jesse Bettencourt, and David K Duvenaud. Neural ordinary differential equa- tions.Advances in neural information processing systems, 31, 2018. 2, 3, 5

    2018

  8. [16]

    Jstasr: Joint size and transparency- aware snow removal algorithm based on modified par- tial convolution and veiling effect removal

    Wei-Ting Chen, Hao-Yu Fang, Jian-Jiun Ding, Cheng-Che Tsai, and Sy-Yen Kuo. Jstasr: Joint size and transparency- aware snow removal algorithm based on modified par- tial convolution and veiling effect removal. InComputer Vision–ECCV 2020: 16th European Conference, Glasgow, UK,...

    2020

  9. [17]

    Learning a sparse transformer network for effective image deraining

    Xiang Chen, Hao Li, Mingqiang Li, and Jinshan Pan. Learning a sparse transformer network for effective image deraining. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 5896– 5905, 2023. 1, 2

    2023

  10. [18]

    Rethinking coarse-to-fine ap- proach in single image deblurring

    Sung-Jin Cho, Seo-Won Ji, Jun-Pyo Hong, Seung-Won Jung, and Sung-Jea Ko. Rethinking coarse-to-fine ap- proach in single image deblurring. InProceedings of the IEEE/CVF international conference on computer vision, pages 4641–4650, 2021. 8

    2021

  11. [19]

    Focal network for image restoration

    Yuning Cui, Wenqi Ren, Xiaochun Cao, and Alois Knoll. Focal network for image restoration. InProceedings of the IEEE/CVF international conference on computer vi- sion, pages 13001–13011, 2023. 5, 6, 7

    2023

  12. [20]

    Inversion by di- rect iteration: An alternative to denoising diffusion for im- age restoration.arXiv preprint arXiv:2303.11435, 2023

    Mauricio Delbracio and Peyman Milanfar. Inversion by di- rect iteration: An alternative to denoising diffusion for im- age restoration.arXiv preprint arXiv:2303.11435, 2023. 2, 3, 4

    2023 arXiv

  13. [21]

    Detail- recovery image deraining via context aggregation networks

    Sen Deng, Mingqiang Wei, Jun Wang, Yidan Feng, Luming Liang, Haoran Xie, Fu Lee Wang, and Meng Wang. Detail- recovery image deraining via context aggregation networks. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 14560–14569, 2020. 1, 2

    2020

  14. [22]

    Multi-scale boosted de- hazing network with dense feature fusion

    Hang Dong, Jinshan Pan, Lei Xiang, Zhe Hu, Xinyi Zhang, Fei Wang, and Ming-Hsuan Yang. Multi-scale boosted de- hazing network with dense feature fusion. InProceedings of the IEEE/CVF conference on computer vision and pat- tern recognition, pages 2157–2167, 2020. 5

    2020

  15. [23]

    Quantization guided jpeg artifact correction

    Max Ehrlich, Larry Davis, Ser-Nam Lim, and Abhinav Shrivastava. Quantization guided jpeg artifact correction. InComputer Vision–ECCV 2020: 16th European Confer- ence, Glasgow, UK, August 23–28, 2020, Proceedings, Part VIII 16, pages 293–309. Springer, 2020. 1, 2, 8

    2020

  16. [24]

    Gener- ative diffusion prior for unified image restoration and en- hancement

    Ben Fei, Zhaoyang Lyu, Liang Pan, Junzhe Zhang, Wei- dong Yang, Tianyue Luo, Bo Zhang, and Bo Dai. Gener- ative diffusion prior for unified image restoration and en- hancement. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 9935– 994...

    2023

  17. [25]

    Generative adversarial networks.Com- munications of the ACM, 63(11):139–144, 2020

    Ian Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio. Generative adversarial networks.Com- munications of the ACM, 63(11):139–144, 2020. 2

    2020

  18. [26]

    Improved training of wasserstein gans.Advances in neural information process- ing systems, 30, 2017

    Ishaan Gulrajani, Faruk Ahmed, Martin Arjovsky, Vincent Dumoulin, and Aaron C Courville. Improved training of wasserstein gans.Advances in neural information process- ing systems, 30, 2017. 2

    2017

  19. [27]

    Image dehazing transformer with transmission-aware 3d position embedding

    Chun-Le Guo, Qixin Yan, Saeed Anwar, Runmin Cong, Wenqi Ren, and Chongyi Li. Image dehazing transformer with transmission-aware 3d position embedding. InPro- ceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 5812–5820, 2022. 1, 2, 5, 6

    2022

  20. [28]

    From sky to the ground: A large-scale bench- mark and simple baseline towards real rain removal

    Yun Guo, Xueyao Xiao, Yi Chang, Shumin Deng, and Luxin Yan. From sky to the ground: A large-scale bench- mark and simple baseline towards real rain removal. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 12097–12107, 2023. 5, 6, 7, 2, 4

    2023

  21. [29]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. InProceed- ings of the IEEE conference on computer vision and pattern recognition, pages 770–778, 2016. 3

    2016

  22. [30]

    Generic image restoration with flow based priors

    Leonhard Helminger, Michael Bernasconi, Abdelaziz Djelouah, Markus Gross, and Christopher Schroers. Generic image restoration with flow based priors. InPro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 334–343, 2021. 2, 5

    2021

  23. [31]

    Denoising dif- fusion probabilistic models.Advances in neural informa- tion processing systems, 33:6840–6851, 2020

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising dif- fusion probabilistic models.Advances in neural informa- tion processing systems, 33:6840–6851, 2020. 2, 5, 3

    2020

  24. [32]

    Selective wavelet attention learning for single im- age deraining.International Journal of Computer Vision, 129(4):1282–1300, 2021

    Huaibo Huang, Aijing Yu, Zhenhua Chai, Ran He, and Tie- niu Tan. Selective wavelet attention learning for single im- age deraining.International Journal of Computer Vision, 129(4):1282–1300, 2021. 6

    2021

  25. [33]

    Arbitrary style transfer in real-time with adaptive instance normalization

    Xun Huang and Serge Belongie. Arbitrary style transfer in real-time with adaptive instance normalization. InPro- ceedings of the IEEE international conference on computer vision, pages 1501–1510, 2017. 3

    2017

  26. [34]

    Image-to-image translation with conditional adver- sarial networks

    Phillip Isola, Jun-Yan Zhu, Tinghui Zhou, and Alexei A Efros. Image-to-image translation with conditional adver- sarial networks. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 1125–1134,

  27. [35]

    Towards flex- ible blind jpeg artifacts removal

    Jiaxi Jiang, Kai Zhang, and Radu Timofte. Towards flex- ible blind jpeg artifacts removal. InProceedings of the IEEE/CVF International Conference on Computer Vision, pages 4997–5006, 2021. 1, 2, 6

    2021

  28. [36]

    Rain- free and residue hand-in-hand: A progressive coupled net- work for real-time image deraining.IEEE Transactions on Image Processing, 30:7404–7418, 2021

    Kui Jiang, Zhongyuan Wang, Peng Yi, Chen Chen, Zheng Wang, Xiao Wang, Junjun Jiang, and Chia-Wen Lin. Rain- free and residue hand-in-hand: A progressive coupled net- work for real-time image deraining.IEEE Transactions on Image Processing, 30:7404–7418, 2021. 5

    2021

  29. [37]

    Edge-based defocus blur estimation with adaptive scale selection.IEEE Trans- actions on Image Processing, 27(3):1126–1137, 2017

    Ali Karaali and Claudio Rosito Jung. Edge-based defocus blur estimation with adaptive scale selection.IEEE Trans- actions on Image Processing, 27(3):1126–1137, 2017. 7

    2017

  30. [38]

    Denoising diffusion restoration models.Advances in Neural Information Processing Systems, 35:23593–23606,

    Bahjat Kawar, Michael Elad, Stefano Ermon, and Jiaming Song. Denoising diffusion restoration models.Advances in Neural Information Processing Systems, 35:23593–23606,

  31. [39]

    Bigcolor: Colorization using a genera- tive color prior for natural images

    Geonung Kim, Kyoungkook Kang, Seongtae Kim, Hway- oon Lee, Sehoon Kim, Jonghyun Kim, Seung-Hwan Baek, and Sunghyun Cho. Bigcolor: Colorization using a genera- tive color prior for natural images. InEuropean Conference on Computer Vision, pages 350–366. Springer, 2022. 1, 2

    2022

  32. [40]

    Auto-encoding variational bayes

    Diederik P Kingma. Auto-encoding variational bayes. arXiv preprint arXiv:1312.6114, 2013. 2

    2013 arXiv

  33. [41]

    Adam: A method for stochastic opti- mization.arXiv preprint arXiv:1412.6980, 2014

    Diederik P Kingma. Adam: A method for stochastic opti- mization.arXiv preprint arXiv:1412.6980, 2014. 5

    2014 arXiv

  34. [42]

    Glow: Generative flow with invertible 1x1 convolutions.Advances in neural information processing systems, 31, 2018

    Durk P Kingma and Prafulla Dhariwal. Glow: Generative flow with invertible 1x1 convolutions.Advances in neural information processing systems, 31, 2018. 2

    2018

  35. [43]

    Estimating mutual information.Physical Review E—Statistical, Nonlinear, and Soft Matter Physics, 69(6): 066138, 2004

    Alexander Kraskov, Harald St ¨ogbauer, and Peter Grass- berger. Estimating mutual information.Physical Review E—Statistical, Nonlinear, and Soft Matter Physics, 69(6): 066138, 2004. 1, 2

    2004

  36. [44]

    Photo-realistic single image super-resolution using a gener- ative adversarial network

    Christian Ledig, Lucas Theis, Ferenc Husz ´ar, Jose Ca- ballero, Andrew Cunningham, Alejandro Acosta, Andrew Aitken, Alykhan Tejani, Johannes Totz, Zehan Wang, et al. Photo-realistic single image super-resolution using a gener- ative adversarial network. InProceedings of the I...

    2017

  37. [45]

    Deep defocus map estimation using domain adapta- tion

    Junyong Lee, Sungkil Lee, Sunghyun Cho, and Seungyong Lee. Deep defocus map estimation using domain adapta- tion. InProceedings of the IEEE/CVF conference on com- puter vision and pattern recognition, pages 12222–12230,

  38. [46]

    Iterative filter adaptive network for single image defocus deblurring

    Junyong Lee, Hyeongseok Son, Jaesung Rim, Sunghyun Cho, and Seungyong Lee. Iterative filter adaptive network for single image defocus deblurring. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 2034–2042, 2021. 7

    2021

  39. [47]

    Aod-net: All-in-one dehazing network

    Boyi Li, Xiulian Peng, Zhangyang Wang, Jizheng Xu, and Dan Feng. Aod-net: All-in-one dehazing network. InPro- ceedings of the IEEE international conference on computer vision, pages 4770–4778, 2017. 5, 6

    2017

  40. [48]

    Srdiff: Single image super-resolution with diffusion probabilistic models

    Haoying Li, Yifan Yang, Meng Chang, Shiqi Chen, Huajun Feng, Zhihai Xu, Qi Li, and Yueting Chen. Srdiff: Single image super-resolution with diffusion probabilistic models. Neurocomputing, 479:47–59, 2022. 2

    2022

  41. [49]

    Heavy rain image restoration: Integrating physics model and conditional adversarial learning

    Ruoteng Li, Loong-Fah Cheong, and Robby T Tan. Heavy rain image restoration: Integrating physics model and conditional adversarial learning. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 1633–1642, 2019. 5, 7, 2, 4

    2019

  42. [50]

    Recurrent squeeze-and-excitation context aggre- gation net for single image deraining

    Xia Li, Jianlong Wu, Zhouchen Lin, Hong Liu, and Hong- bin Zha. Recurrent squeeze-and-excitation context aggre- gation net for single image deraining. InProceedings of the European conference on computer vision (ECCV), pages 254–269, 2018. 5

    2018

  43. [51]

    Diffbir: Towards blind image restoration with generative diffusion prior.arXiv preprint arXiv:2308.15070, 2023

    Xinqi Lin, Jingwen He, Ziyan Chen, Zhaoyang Lyu, Bo Dai, Fanghua Yu, Wanli Ouyang, Yu Qiao, and Chao Dong. Diffbir: Towards blind image restoration with generative diffusion prior.arXiv preprint arXiv:2308.15070, 2023. 1, 2

    2023 arXiv

  44. [52]

    Catch missing details: Image reconstruction with frequency augmented variational autoencoder

    Xinmiao Lin, Yikang Li, Jenhao Hsiao, Chiuman Ho, and Yu Kong. Catch missing details: Image reconstruction with frequency augmented variational autoencoder. InProceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 1736–1745, 2023. 2

    2023

  45. [53]

    Unsupervised image denoising in real-world scenarios via self-collaboration parallel generative adversarial branches

    Xin Lin, Chao Ren, Xiao Liu, Jie Huang, and Yinjie Lei. Unsupervised image denoising in real-world scenarios via self-collaboration parallel generative adversarial branches. InProceedings of the IEEE/CVF International Conference on Computer Vision, pages 12642–12652, 2023. 1, 2

    2023

  46. [54]

    Flow matching for generative modeling.arXiv preprint arXiv:2210.02747, 2022

    Yaron Lipman, Ricky TQ Chen, Heli Ben-Hamu, Maxim- ilian Nickel, and Matt Le. Flow matching for generative modeling.arXiv preprint arXiv:2210.02747, 2022. 2, 3, 4, 5

    2022 arXiv

  47. [55]

    I2sb: Image-to-image schr ¨odinger bridge.arXiv preprint arXiv:2302.05872, 2023

    Guan-Horng Liu, Arash Vahdat, De-An Huang, Evange- los A Theodorou, Weili Nie, and Anima Anandkumar. I2sb: Image-to-image schr ¨odinger bridge.arXiv preprint arXiv:2302.05872, 2023. 2

    2023 arXiv

  48. [56]

    Structure matters: Tackling the semantic dis- crepancy in diffusion models for image inpainting

    Haipeng Liu, Yang Wang, Biao Qian, Meng Wang, and Yong Rui. Structure matters: Tackling the semantic dis- crepancy in diffusion models for image inpainting. InPro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 8038–8047, 2024. 2

    2024

  49. [57]

    Residual denoising diffu- sion models

    Jiawei Liu, Qiang Wang, Huijie Fan, Yinong Wang, Yan- dong Tang, and Liangqiong Qu. Residual denoising diffu- sion models. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 2773– 2783, 2024. 2, 3

    2024

  50. [58]

    Flow straight and fast: Learning to generate and transfer data with rectified flow.arXiv preprint arXiv:2209.03003, 2022

    Xingchao Liu, Chengyue Gong, and Qiang Liu. Flow straight and fast: Learning to generate and transfer data with rectified flow.arXiv preprint arXiv:2209.03003, 2022. 4, 5

    2022 arXiv

  51. [59]

    Diff-plugin: Revitalizing details for diffusion-based low-level tasks

    Yuhao Liu, Zhanghan Ke, Fang Liu, Nanxuan Zhao, and Rynson WH Lau. Diff-plugin: Revitalizing details for diffusion-based low-level tasks. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 4197–4208, 2024. 2

    2024

  52. [60]

    Desnownet: Context-aware deep network for snow removal.IEEE Transactions on Image Processing, 27(6): 3064–3073, 2018

    Yun-Fu Liu, Da-Wei Jaw, Shih-Chia Huang, and Jenq-Neng Hwang. Desnownet: Context-aware deep network for snow removal.IEEE Transactions on Image Processing, 27(6): 3064–3073, 2018. 1, 2, 5, 4

    2018

  53. [61]

    Decoupled weight decay regularization.arXiv preprint arXiv:1711.05101, 2017

    I Loshchilov. Decoupled weight decay regularization.arXiv preprint arXiv:1711.05101, 2017. 3

    2017 arXiv

  54. [62]

    Sgdr: Stochastic gradient descent with warm restarts.arXiv preprint arXiv:1608.03983, 2016

    Ilya Loshchilov and Frank Hutter. Sgdr: Stochastic gradient descent with warm restarts.arXiv preprint arXiv:1608.03983, 2016. 3

    2016 arXiv

  55. [63]

    Normalizing flow as a flexi- ble fidelity objective for photo-realistic super-resolution

    Andreas Lugmayr, Martin Danelljan, Fisher Yu, Luc Van Gool, and Radu Timofte. Normalizing flow as a flexi- ble fidelity objective for photo-realistic super-resolution. In Proceedings of the IEEE/CVF winter conference on appli- cations of computer vision, pages 1756–1765, 2022. 2, 5

    2022

  56. [64]

    Image restoration with mean-reverting stochastic differential equations.arXiv preprint arXiv:2301.11699, 2023

    Ziwei Luo, Fredrik K Gustafsson, Zheng Zhao, Jens Sj¨olund, and Thomas B Sch ¨on. Image restoration with mean-reverting stochastic differential equations.arXiv preprint arXiv:2301.11699, 2023. 2

    2023 arXiv

  57. [65]

    Refusion: Enabling large- size realistic image restoration with latent-space diffusion models

    Ziwei Luo, Fredrik K Gustafsson, Zheng Zhao, Jens Sj¨olund, and Thomas B Sch ¨on. Refusion: Enabling large- size realistic image restoration with latent-space diffusion models. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 1680–1691,

  58. [66]

    Sen- sitivity decouple learning for image compression artifacts reduction.IEEE Transactions on Image Processing, 2024

    Li Ma, Yifan Zhao, Peixi Peng, and Yonghong Tian. Sen- sitivity decouple learning for image compression artifacts reduction.IEEE Transactions on Image Processing, 2024. 8

    2024

  59. [67]

    Least squares gen- erative adversarial networks

    Xudong Mao, Qing Li, Haoran Xie, Raymond YK Lau, Zhen Wang, and Stephen Paul Smolley. Least squares gen- erative adversarial networks. InProceedings of the IEEE international conference on computer vision, pages 2794– 2802, 2017. 2

    2017

  60. [68]

    Deep residual fourier transformation for single im- age deblurring.arXiv preprint arXiv:2111.11745, 2(3):5,

    Xintian Mao, Yiming Liu, Wei Shen, Qingli Li, and Yan Wang. Deep residual fourier transformation for single im- age deblurring.arXiv preprint arXiv:2111.11745, 2(3):5,

    arXiv

  61. [69]

    A database of human segmented natural images and its application to evaluating segmentation algorithms and measuring ecological statistics

    David Martin, Charless Fowlkes, Doron Tal, and Jitendra Malik. A database of human segmented natural images and its application to evaluating segmentation algorithms and measuring ecological statistics. InProceedings eighth IEEE international conference on computer vision. ICC...

    2001

  62. [70]

    Conditional generative adversarial nets

    Mehdi Mirza. Conditional generative adversarial nets. arXiv preprint arXiv:1411.1784, 2014. 2

    2014 arXiv

  63. [71]

    Noisier2noise: Learning to denoise from unpaired noisy data

    Nick Moran, Dan Schmidt, Yu Zhong, and Patrick Coady. Noisier2noise: Learning to denoise from unpaired noisy data. InProceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition, pages 12064–12072,

  64. [72]

    Dynamic at- tentive graph learning for image restoration

    Chong Mou, Jian Zhang, and Zhuoyuan Wu. Dynamic at- tentive graph learning for image restoration. InProceedings of the IEEE/CVF international conference on computer vi- sion, pages 4328–4337, 2021. 6

    2021

  65. [73]

    T2i-adapter: Learn- ing adapters to dig out more controllable ability for text-to- image diffusion models

    Chong Mou, Xintao Wang, Liangbin Xie, Yanze Wu, Jian Zhang, Zhongang Qi, and Ying Shan. T2i-adapter: Learn- ing adapters to dig out more controllable ability for text-to- image diffusion models. InProceedings of the AAAI Con- ference on Artificial Intelligence, pages 4296–4304...

    2024

  66. [74]

    Cdpmsr: Condi- tional diffusion probabilistic models for single image super- resolution.arXiv preprint arXiv:2302.12831, 2023

    Axi Niu, Kang Zhang, Trung X Pham, Jinqiu Sun, Yu Zhu, In So Kweon, and Yanning Zhang. Cdpmsr: Condi- tional diffusion probabilistic models for single image super- resolution.arXiv preprint arXiv:2302.12831, 2023. 3

    2023 arXiv

  67. [75]

    Ot-flow: Fast and accurate continuous normal- izing flows via optimal transport

    Derek Onken, Samy Wu Fung, Xingjian Li, and Lars Ruthotto. Ot-flow: Fast and accurate continuous normal- izing flows via optimal transport. InProceedings of the AAAI Conference on Artificial Intelligence, pages 9223– 9232, 2021. 3

    2021

  68. [76]

    Restoring vision in adverse weather conditions with patch-based denoising diffusion models.IEEE Transactions on Pattern Analysis and Machine Intelligence, 45(8):10346–10357, 2023

    Ozan ¨Ozdenizci and Robert Legenstein. Restoring vision in adverse weather conditions with patch-based denoising diffusion models.IEEE Transactions on Pattern Analysis and Machine Intelligence, 45(8):10346–10357, 2023. 1, 2, 5, 8, 4

    2023

  69. [77]

    Exploiting deep genera- tive prior for versatile image restoration and manipulation

    Xingang Pan, Xiaohang Zhan, Bo Dai, Dahua Lin, Chen Change Loy, and Ping Luo. Exploiting deep genera- tive prior for versatile image restoration and manipulation. IEEE Transactions on Pattern Analysis and Machine Intel- ligence, 44(11):7474–7489, 2021. 1, 2

    2021

  70. [78]

    Normalizing flows for probabilistic modeling and infer- ence.Journal of Machine Learning Research, 22(57):1–64,

    George Papamakarios, Eric Nalisnick, Danilo Jimenez Rezende, Shakir Mohamed, and Balaji Lakshminarayanan. Normalizing flows for probabilistic modeling and infer- ence.Journal of Machine Learning Research, 22(57):1–64,

  71. [79]

    Ffa-net: Feature fusion attention network for single image dehazing

    Xu Qin, Zhilin Wang, Yuanchao Bai, Xiaodong Xie, and Huizhu Jia. Ffa-net: Feature fusion attention network for single image dehazing. InProceedings of the AAAI confer- ence on artificial intelligence, pages 11908–11915, 2020. 5

    2020

  72. [80]

    Mb-taylorformer: Multi-branch efficient transformer expanded by taylor formula for im- age dehazing

    Yuwei Qiu, Kaihao Zhang, Chenxi Wang, Wenhan Luo, Hongdong Li, and Zhi Jin. Mb-taylorformer: Multi-branch efficient transformer expanded by taylor formula for im- age dehazing. InProceedings of the IEEE/CVF Inter- national Conference on Computer Vision, pages 12802– 12813, 2023. 5

    2023

  73. [81]

    Adaptive consistency prior based deep network for image denoising

    Chao Ren, Xiaohai He, Chuncheng Wang, and Zhibo Zhao. Adaptive consistency prior based deep network for image denoising. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 8596–8606,

  74. [82]

    Progressive image deraining net- works: A better and simpler baseline

    Dongwei Ren, Wangmeng Zuo, Qinghua Hu, Pengfei Zhu, and Deyu Meng. Progressive image deraining net- works: A better and simpler baseline. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 3937–3946, 2019. 6

    2019

  75. [83]

    U- net: Convolutional networks for biomedical image segmen- tation

    Olaf Ronneberger, Philipp Fischer, and Thomas Brox. U- net: Convolutional networks for biomedical image segmen- tation. InMedical image computing and computer-assisted intervention–MICCAI 2015: 18th international conference, Munich, Germany, October 5-9, 2015, proceedings, par...

    2015

  76. [84]

    Learning to deblur using light field generated and real de- focus images

    Lingyan Ruan, Bin Chen, Jizhou Li, and Miuling Lam. Learning to deblur using light field generated and real de- focus images. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 16304– 16313, 2022. 7

    2022

  77. [85]

    Improved techniques for training gans.Advances in neural information process- ing systems, 29, 2016

    Tim Salimans, Ian Goodfellow, Wojciech Zaremba, Vicki Cheung, Alec Radford, and Xi Chen. Improved techniques for training gans.Advances in neural information process- ing systems, 29, 2016. 2

    2016

  78. [86]

    Resdiff: Combining cnn and diffusion model for image super-resolution

    Shuyao Shang, Zhengyang Shan, Guangxing Liu, LunQian Wang, XingHua Wang, Zekai Zhang, and Jinglin Zhang. Resdiff: Combining cnn and diffusion model for image super-resolution. InProceedings of the AAAI Conference on Artificial Intelligence, pages 8975–8983, 2024. 2

    2024

  79. [87]

    Live image quality assessment database release 2.http://live

    H Sheikh. Live image quality assessment database release 2.http://live. ece. utexas. edu/research/quality, 2005. 5, 8, 3

    2005

  80. [88]

    Just noticeable defocus blur detection and estimation

    Jianping Shi, Li Xu, and Jiaya Jia. Just noticeable defocus blur detection and estimation. InProceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 657–665, 2015. 7

    2015

  81. [89]

    Resfusion: Denoising diffusion probabilistic models for image restoration based on prior residual noise

    Zhenning Shi, Haoshuai Zheng, Chen Xu, Changsheng Dong, Bin Pan, Xueshuo Xie, Along He, Tao Li, and Huazhu Fu. Resfusion: Denoising diffusion probabilistic models for image restoration based on prior residual noise. arXiv e-prints, pages arXiv–2311, 2023. 2

    2023

  82. [90]

    Variational deep image restoration.IEEE Transactions on Image Processing, 31: 4363–4376, 2022

    Jae Woong Soh and Nam Ik Cho. Variational deep image restoration.IEEE Transactions on Image Processing, 31: 4363–4376, 2022. 2

    2022

  83. [91]

    Deep unsupervised learning using nonequilibrium thermodynamics

    Jascha Sohl-Dickstein, Eric Weiss, Niru Maheswaranathan, and Surya Ganguli. Deep unsupervised learning using nonequilibrium thermodynamics. InInternational confer- ence on machine learning, pages 2256–2265. PMLR, 2015. 2

    2015

  84. [92]

    Single image defocus deblurring using kernel-sharing parallel atrous convolutions

    Hyeongseok Son, Junyong Lee, Sunghyun Cho, and Se- ungyong Lee. Single image defocus deblurring using kernel-sharing parallel atrous convolutions. InProceedings of the IEEE/CVF International Conference on Computer Vision, pages 2642–2650, 2021. 7

    2021

  85. [93]

    Score- based generative modeling through stochastic differential equations.arXiv preprint arXiv:2011.13456, 2020

    Yang Song, Jascha Sohl-Dickstein, Diederik P Kingma, Abhishek Kumar, Stefano Ermon, and Ben Poole. Score- based generative modeling through stochastic differential equations.arXiv preprint arXiv:2011.13456, 2020. 2

    2011 arXiv

  86. [94]

    Transweather: Transformer-based restoration of im- ages degraded by adverse weather conditions

    Jeya Maria Jose Valanarasu, Rajeev Yasarla, and Vishal M Patel. Transweather: Transformer-based restoration of im- ages degraded by adverse weather conditions. InProceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 2353–2363, 2022. 6

    2022

  87. [95]

    Spatial attentive single-image deraining with a high quality real rain dataset

    Tianyu Wang, Xin Yang, Ke Xu, Shaozhe Chen, Qiang Zhang, and Rynson WH Lau. Spatial attentive single-image deraining with a high quality real rain dataset. InProceed- ings of the IEEE/CVF conference on computer vision and pattern recognition, pages 12270–12279, 2019. 5, 6

    2019

  88. [96]

    Esrgan: En- hanced super-resolution generative adversarial networks

    Xintao Wang, Ke Yu, Shixiang Wu, Jinjin Gu, Yihao Liu, Chao Dong, Yu Qiao, and Chen Change Loy. Esrgan: En- hanced super-resolution generative adversarial networks. In Proceedings of the European conference on computer vi- sion (ECCV) workshops, pages 0–0, 2018. 1, 2

    2018

  89. [97]

    To- wards real-world blind face restoration with generative fa- cial prior

    Xintao Wang, Yu Li, Honglun Zhang, and Ying Shan. To- wards real-world blind face restoration with generative fa- cial prior. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 9168–9178, 2021

    2021

  90. [98]

    Real-esrgan: Training real-world blind super-resolution with pure synthetic data

    Xintao Wang, Liangbin Xie, Chao Dong, and Ying Shan. Real-esrgan: Training real-world blind super-resolution with pure synthetic data. InProceedings of the IEEE/CVF international conference on computer vision, pages 1905– 1914, 2021. 2

    1905

  91. [99]

    Low-light image enhancement with normalizing flow

    Yufei Wang, Renjie Wan, Wenhan Yang, Haoliang Li, Lap- Pui Chau, and Alex Kot. Low-light image enhancement with normalizing flow. InProceedings of the AAAI con- ference on artificial intelligence, pages 2604–2612, 2022. 2

    2022

  92. [100]

    Zero-shot im- age restoration using denoising diffusion null-space model

    Yinhuai Wang, Jiwen Yu, and Jian Zhang. Zero-shot im- age restoration using denoising diffusion null-space model. arXiv preprint arXiv:2212.00490, 2022. 1, 2

    2022 arXiv

  93. [101]

    Smartas- sign: Learning a smart knowledge assignment strategy for deraining and desnowing

    Yinglong Wang, Chao Ma, and Jianzhuang Liu. Smartas- sign: Learning a smart knowledge assignment strategy for deraining and desnowing. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 3677–3686, 2023. 1, 2

    2023

  94. [102]

    Image quality assessment: from error visibility to structural similarity.IEEE transactions on image pro- cessing, 13(4):600–612, 2004

    Zhou Wang, Alan C Bovik, Hamid R Sheikh, and Eero P Simoncelli. Image quality assessment: from error visibility to structural similarity.IEEE transactions on image pro- cessing, 13(4):600–612, 2004. 5

    2004

  95. [103]

    Con- trastive learning for compact single image dehazing

    Haiyan Wu, Yanyun Qu, Shaohui Lin, Jian Zhou, Ruizhi Qiao, Zhizhong Zhang, Yuan Xie, and Lizhuang Ma. Con- trastive learning for compact single image dehazing. InPro- ceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 10551–10560, 2021. 5, 6

    2021

  96. [104]

    Diffir: Efficient diffusion model for image restoration

    Bin Xia, Yulun Zhang, Shiyin Wang, Yitong Wang, Xing- long Wu, Yapeng Tian, Wenming Yang, and Luc Van Gool. Diffir: Efficient diffusion model for image restoration. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 13095–13105, 2023. 2

    2023

  97. [105]

    Tackling the generative learning trilemma with denoising diffusion gans.arXiv preprint arXiv:2112.07804, 2021

    Zhisheng Xiao, Karsten Kreis, and Arash Vahdat. Tackling the generative learning trilemma with denoising diffusion gans.arXiv preprint arXiv:2112.07804, 2021. 2

    2021 arXiv

  98. [106]

    Understanding and improving layer nor- malization.Advances in neural information processing sys- tems, 32, 2019

    Jingjing Xu, Xu Sun, Zhiyuan Zhang, Guangxiang Zhao, and Junyang Lin. Understanding and improving layer nor- malization.Advances in neural information processing sys- tems, 32, 2019. 5

    2019

  99. [107]

    Scale-free single image deraining via visibility- enhanced recurrent wavelet learning.IEEE Transactions on Image Processing, 28(6):2948–2961, 2019

    Wenhan Yang, Jiaying Liu, Shuai Yang, and Zongming Guo. Scale-free single image deraining via visibility- enhanced recurrent wavelet learning.IEEE Transactions on Image Processing, 28(6):2948–2961, 2019. 8

    2019

  100. [108]

    Perceiving and modeling density for image dehazing

    Tian Ye, Yunchen Zhang, Mingchao Jiang, Liang Chen, Yun Liu, Sixiang Chen, and Erkang Chen. Perceiving and modeling density for image dehazing. InEuropean confer- ence on computer vision, pages 130–145. Springer, 2022. 5, 6

    2022

  101. [109]

    Learning diffusion texture pri- ors for image restoration

    Tian Ye, Sixiang Chen, Wenhao Chai, Zhaohu Xing, Jing Qin, Ge Lin, and Lei Zhu. Learning diffusion texture pri- ors for image restoration. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 2524–2534, 2024. 5, 7

    2024

  102. [110]

    Scaling up to excellence: Practicing model scaling for photo-realistic image restoration in the wild

    Fanghua Yu, Jinjin Gu, Zheyuan Li, Jinfan Hu, Xiang- tao Kong, Xintao Wang, Jingwen He, Yu Qiao, and Chao Dong. Scaling up to excellence: Practicing model scaling for photo-realistic image restoration in the wild. InPro- ceedings of the IEEE/CVF Conference on Computer Vision a...

    2024

  103. [111]

    Dual adversarial network: Toward real-world noise removal and noise generation

    Zongsheng Yue, Qian Zhao, Lei Zhang, and Deyu Meng. Dual adversarial network: Toward real-world noise removal and noise generation. InComputer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part X 16, pages 41–58. Springer, 2020. 6

    2020

  104. [112]

    Resshift: Efficient diffusion model for image super- resolution by residual shifting.Advances in Neural Infor- mation Processing Systems, 36, 2024

    Zongsheng Yue, Jianyi Wang, and Chen Change Loy. Resshift: Efficient diffusion model for image super- resolution by residual shifting.Advances in Neural Infor- mation Processing Systems, 36, 2024. 2

    2024

  105. [113]

    Learning enriched features for real image restoration and enhancement

    Syed Waqas Zamir, Aditya Arora, Salman Khan, Munawar Hayat, Fahad Shahbaz Khan, Ming-Hsuan Yang, and Ling Shao. Learning enriched features for real image restoration and enhancement. InComputer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Procee...

    2020

  106. [114]

    Multi-stage progressive image restoration

    Syed Waqas Zamir, Aditya Arora, Salman Khan, Munawar Hayat, Fahad Shahbaz Khan, Ming-Hsuan Yang, and Ling Shao. Multi-stage progressive image restoration. InPro- ceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 14821–14831, 2021. 5, 6

    2021

  107. [115]

    Restormer: Efficient transformer for high-resolution image restoration

    Syed Waqas Zamir, Aditya Arora, Salman Khan, Mu- nawar Hayat, Fahad Shahbaz Khan, and Ming-Hsuan Yang. Restormer: Efficient transformer for high-resolution image restoration. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 5728–5739,

  108. [116]

    Learning enriched features for fast image restoration and enhancement.IEEE transactions on pattern analysis and machine intelligence, 45(2):1934–1948, 2022

    Syed Waqas Zamir, Aditya Arora, Salman Khan, Munawar Hayat, Fahad Shahbaz Khan, Ming-Hsuan Yang, and Ling Shao. Learning enriched features for fast image restoration and enhancement.IEEE transactions on pattern analysis and machine intelligence, 45(2):1934–1948, 2022. 6

    1934

  109. [117]

    Mm-bsn: Self-supervised image denoising for real- world with multi-mask based on blind-spot network

    Dan Zhang, Fangfang Zhou, Yuwen Jiang, and Zhengming Fu. Mm-bsn: Self-supervised image denoising for real- world with multi-mask based on blind-spot network. In Proceedings of the IEEE/CVF Conference on Computer Vi- sion and Pattern Recognition, pages 4189–4198, 2023. 1, 2

    2023

  110. [118]

    Accurate image restora- tion with attention retractable transformer.arXiv preprint arXiv:2210.01427, 2022

    Jiale Zhang, Yulun Zhang, Jinjin Gu, Yongbing Zhang, Linghe Kong, and Xin Yuan. Accurate image restora- tion with attention retractable transformer.arXiv preprint arXiv:2210.01427, 2022. 6

    2022 arXiv

  111. [119]

    Xformer: Hybrid x- shaped transformer for image denoising.arXiv preprint arXiv:2303.06440, 2023

    Jiale Zhang, Yulun Zhang, Jinjin Gu, Jiahua Dong, Linghe Kong, and Xiaokang Yang. Xformer: Hybrid x- shaped transformer for image denoising.arXiv preprint arXiv:2303.06440, 2023. 6

    2023 arXiv

  112. [120]

    The unreasonable effectiveness of deep features as a perceptual metric

    Richard Zhang, Phillip Isola, Alexei A Efros, Eli Shecht- man, and Oliver Wang. The unreasonable effectiveness of deep features as a perceptual metric. InProceedings of the IEEE conference on computer vision and pattern recogni- tion, pages 586–595, 2018. 2, 5

    2018

  113. [121]

    Diffusion-based blind text image super-resolution

    Yuzhe Zhang, Jiawei Zhang, Hao Li, Zhouxia Wang, Luwei Hou, Dongqing Zou, and Liheng Bian. Diffusion-based blind text image super-resolution. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 25827–25836, 2024. 2

    2024

  114. [122]

    Loss functions for image restoration with neural networks.IEEE Transactions on computational imaging, 3(1):47–57, 2016

    Hang Zhao, Orazio Gallo, Iuri Frosio, and Jan Kautz. Loss functions for image restoration with neural networks.IEEE Transactions on computational imaging, 3(1):47–57, 2016. 2

    2016

  115. [123]

    Learn from unpaired data for image restoration: A variational bayes approach.IEEE Transactions on Pat- tern Analysis and Machine Intelligence, 45(5):5889–5903,

    Dihan Zheng, Xiaowen Zhang, Kaisheng Ma, and Cheng- long Bao. Learn from unpaired data for image restoration: A variational bayes approach.IEEE Transactions on Pat- tern Analysis and Machine Intelligence, 45(5):5889–5903,

  116. [124]

    Learning to dehaze with polarization.Advances in neural information processing systems, 34:11487–11500,

    Chu Zhou, Minggui Teng, Yufei Han, Chao Xu, and Boxin Shi. Learning to dehaze with polarization.Advances in neural information processing systems, 34:11487–11500,

  117. [125]

    Unpaired image-to-image translation using cycle- consistent adversarial networks

    Jun-Yan Zhu, Taesung Park, Phillip Isola, and Alexei A Efros. Unpaired image-to-image translation using cycle- consistent adversarial networks. InProceedings of the IEEE international conference on computer vision, pages 2223– 2232, 2017. 5

    2017

  118. [126]

    Learning weather-general and weather-specific features for image restoration under multiple adverse weather conditions

    Yurui Zhu, Tianyu Wang, Xueyang Fu, Xuanyu Yang, Xin Guo, Jifeng Dai, Yu Qiao, and Xiaowei Hu. Learning weather-general and weather-specific features for image restoration under multiple adverse weather conditions. In Proceedings of the IEEE/CVF conference on computer vi- sion...

    2023

  119. [127]

    Proof of Proposition 1 Here, we restate Proposition 1 from Sec. 3.1. Proposition 1.Given a random process{z t|0≤t≤1} defined by ∂zt ∂t =v(z t,t); 0≤t≤1,(12) wherev∈C 1 is a velocity field, denote the mutual infor- mation asMI(·,·), then for any reference random variable rand a...

  120. [128]

    3.2, we propose the following entropy-preserving degradation schedule: αx t = 1−t, σ x t = 1−α x t (21) σy t =β·(1−t+β) −1, σ y t = 1−α y t,(22) whereβ= 10is a hyperparameter

    Derivation of Entropy-Preserving Degrada- tion Schedule In Sec. 3.2, we propose the following entropy-preserving degradation schedule: αx t = 1−t, σ x t = 1−α x t (21) σy t =β·(1−t+β) −1, σ y t = 1−α y t,(22) whereβ= 10is a hyperparameter. The intuition is that the entropy sho...

  121. [129]

    Datasets We evaluate the performance of image restoration methods on five major image restoration tasks, including desnowing, draining, dehazing, denoising, and JPEG compression arti- fact removal, using synthetic and real-world datasets. De- tails of the datasets are given be...

  122. [130]

    which contains 29 testing images, and BSD500 [6] which contains 500 testing images

  123. [131]

    Model architecture

    Implementation Details Figure 7. Model architecture. All experiments adopt the same U-Net architecture from

  124. [132]

    The input to the U-Net is thext start- ing fromx 1 as the LQ image

    as the backbone. The input to the U-Net is thext start- ing fromx 1 as the LQ image. The output of the U-Net is the velocityvθ(xt,yt,t). We remove the class-label condi- tioning and condition the model ony t via an adapter [73] as illustrated in Fig. 7. The adapter processesy ...

  125. [133]

    generation paths

    Discussion of Different Approaches Image restoration is modeled as paired image-to-image translation from LQs to HQs. Historically image restora- tion methods areone-stepmodels that map LQs to HQs with one single step of inference. Diffusion-based mod- els, however, aremulti-s...

  126. [134]

    However, our ResFlow can generate high qual- ity restored in as few as two or even one step

    Addition Experimental Results Computational costs.Diffusion models are notoriously slow because they require dozens or even hundreds of infer- ence steps. However, our ResFlow can generate high qual- ity restored in as few as two or even one step. Compared with diffusion-based...

  127. [135]

    information difference

    is shown in Fig. 10, our method significantly outper- forms Restormer [115] perceptually. Figure 11 shows the impact of auxiliary variables on the generated results. After optimizing Equations (9) and (10), ResFlow learns a deterministic coupling from the joint dis- tribution ...

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.