Pith. sign in

REVIEW 4 major objections 6 minor 71 references

Frequency Domain-Based Diffusion Model for Unpaired Image Dehazing

T0 review · 4 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read FrDiff removes haze by having a diffusion model reconstruct the amplitude spectrum of clear images, and reports higher unpaired dehazing results on synthetic and real-world benchmarks than the compared baselines.

desk verdict FrDiff is a novel take on unpaired dehazing, but its SOTS-Indoor numbers contradict each other by 9 dB and must be reconciled before the claims are trustworthy. read the letter →

arxiv 2507.01275 v1 pith:7KDPMADD submitted 2025-07-02 cs.CV

Chengxu Liu , Lu Qi , Jinshan Pan , Xueming Qian , Ming-Hsuan Yang This is my paper · ORCID
classification cs.CV
keywords unpairedimagedehazingdiffusionmodelfrequencydomainamplitudespectrumphasecorrectionrestorationcontrastivelearningresidualencoder
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

The paper presents FrDiff, described as the first diffusion model for unpaired image dehazing. Its claim is that haze degrades images mainly through the amplitude spectrum, the frequency component that carries illumination contrast, while the phase spectrum carries texture, so an unpaired dehazing model can remove haze by reconstructing only the amplitude spectrum of the clear domain. To do this it introduces an Amplitude Residual Encoder that aligns the global mean and standard deviation of a hazy image's amplitude to those of an unpaired clear image, and a Phase Correction Module that refines phase to suppress artifacts. On the RESIDE benchmarks it reports 27.43 dB PSNR and 0.957 SSIM on SOTS-Indoor and 22.75 dB and 0.914 SSIM on SOTS-Outdoor with 8.76M parameters and 54.56 GFLOPs. If the claim holds, diffusion models become a practical tool for unpaired restoration because the generative burden is moved from full images to a small frequency-domain residual.

What carries the argument

The load-bearing object is the amplitude residual $z = A'_h - A_h$, where $A_h$ is the amplitude spectrum of the hazy image and $A'_h = (\sigma_c/\sigma_h)(A_h - \mu_h) + \mu_c$ is the hazy amplitude after aligning its global mean and standard deviation to an unpaired clear amplitude spectrum. The Amplitude Residual Encoder computes $z$ in closed form with no learnable parameters; a conditional diffusion model with eight denoising steps generates $\hat{z}$ from Gaussian noise conditioned on the hazy amplitude; frequency compensation layers inject $\hat{z}$ into the dehazing network at multiple scales; and the Phase Correction Module uses a softmax-weighted convolution on the phase spectrum to remove artifacts. The residual formulation keeps the diffusion target small and illumination-only, which is why the method can work with few iterations.

What would settle it

Create synthetic hazy images with spatially varying transmission maps, such as thick haze near the top and thin haze near the bottom, and compare FrDiff's output to ground truth at a fixed average haze density; if its PSNR drops much more on these nonuniform images than on uniformly hazy images, the global amplitude-statistics assumption is the bottleneck.

Watch

Extended reading notes

Core claim

The central discovery is that unpaired image dehazing can be posed as a frequency-domain reconstruction task: instead of translating a hazy image into a clear image directly, the model generates the amplitude residual $z$ that closes the distribution gap between hazy and clear amplitudes. The paper's evidence is that swapping in a clear image's amplitude spectrum brings the dark-channel statistics of a hazy image close to those of true clear images, that the residual estimated by the Amplitude Residual Encoder provides supervision for diffusion training without extra parameters, and that the full FrDiff system outperforms the compared unpaired dehazing methods on SOTS-Indoor, SOTS-Outdoor, HSTS-Synth, and real-world benchmarks. The system does this with only eight denoising steps and a 0.07M-parameter denoiser attached to an 8.69M-parameter dehazing network.

Load-bearing premise

The method assumes that haze is captured by a global shift in the amplitude spectrum, so aligning the overall mean and standard deviation of the hazy amplitude to an unpaired clear amplitude produces a residual that fully describes the haze; if real haze varies spatially across the image, this residual will be miscalibrated no matter how strong the diffusion model is.

Editorial extensions

If this is right

  • Unpaired dehazing can be accomplished by a diffusion model producing only an amplitude residual, requiring about eight denoising steps rather than full image generation.
  • The reported SOTS-Indoor and SOTS-Outdoor results imply the approach beats cycle-consistency and contrastive-learning unpaired baselines at a lower parameter count of 8.76M.
  • The closed-form amplitude alignment supplies diffusion supervision without learnable parameters, so the training signal is tied directly to haze-related amplitude statistics.
  • The Phase Correction Module contributes a small but consistent gain, indicating that phase refinement is needed to remove artifacts left by amplitude-only reconstruction.
  • Ablation results attribute the largest single drop of 2.77 dB to removing the diffusion model, making the diffusion reconstruction the main driver of the reported performance.

Reading between the lines

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

  • An implication the paper leaves implicit is that the same two-stage diffusion-on-residual recipe could transfer to other unpaired restoration problems whose degradation is largely global, such as low-light enhancement, color-cast correction, or uniform blur.
  • Because the model without the diffusion component still aligns amplitudes but scores 24.66 dB, the closed-form alignment alone is not enough; the learned denoiser is what turns the residual into a working dehaze, suggesting a fair comparison against a non-diffusion regressor trained on the same residual.
  • A testable extension is spatially adaptive amplitude alignment; since the paper flags spatially varying haze as a limitation, showing that local or depth-conditional alignment improves real-world results would confirm that the global-statistics assumption, not network capacity, is the bottleneck.
  • The reported gain saturates as denoising steps grow from eight to thirty-two, hinting that a distilled single-step version could retain most of the benefit at a fraction of the FLOPs; the paper mentions one-step distillation as future acceleration.
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

4 major / 6 minor

Summary. The paper proposes FrDiff, an unpaired image dehazing method built around a frequency-domain diffusion model. The method first computes an amplitude residual z by aligning the global mean and standard deviation of the hazy amplitude spectrum to that of an unpaired clear image (Eq. 3). This residual is used both as a training target and as input to a Frequency Compensation Layer in a NAFNet-based dehazing network, while a Phase Correction Module refines the phase spectrum. Training is split into two stages: stage one trains the dehazing network with the ARE residual directly; stage two jointly trains a small denoising network and the dehazing network with an L1 loss on the final reverse-diffusion sample (Eq. 11). Experiments on SOTS-Indoor/Outdoor, HSTS-Synth, I-HAZE, O-HAZE, and real-world sets are reported, with the main table claiming 27.43 dB PSNR on SOTS-Indoor.

Significance. If the reported numbers are reliable, the paper would be a meaningful contribution to unpaired dehazing: it introduces a frequency-domain perspective, shows that a lightweight diffusion model (8.76M parameters, 54.56 GFLOPs at T=8) can be conditioned on amplitude spectra, and provides detailed ablations, algorithm pseudocode, and supplementary dark-channel statistics supporting the amplitude-based motivation. The central claim, however, is currently not internally reproducible. The main table reports 27.43 dB on SOTS-Indoor, while the supplementary hyperparameter table reports 36.54 dB for the same model and default settings, a 9.11 dB inconsistency. The absence of code, weights, seeds, and error bars further prevents verification. The conceptual idea is plausible and the paper is generally clearly written, but the empirical foundation of the headline result needs to be resolved before the contribution can be accepted.

major comments (4)
  1. [Table 1 vs. Supplementary Table A2] The main paper reports 27.43 dB PSNR for FrDiff on SOTS-Indoor (Table 1; also the 'Full model' row of Table 3), while Supplementary Table A2 reports 36.54 dB for the same model with the same default hyperparameters (lambda_GAN = lambda_PatchNCE = lambda_diff = 1). The sweep in Table A2 is essentially flat (36.31-36.58 dB), so hyperparameter sensitivity cannot explain a 9.11 dB gap, and no text statement indicates a different split, resolution, or metric implementation for the two tables. The manuscript therefore contains two mutually inconsistent measurements of its central result, and the headline claim that FrDiff outperforms prior unpaired dehazing methods is not internally reproducible. Please correct the discrepancy, specify exactly which protocol produced each table, and provide the code or weights needed to verify the reported numbers.
  2. [Sec. 4.4, Table 3] The 'w/o DM' ablation uses the ARE residual directly, and the full model uses a DM that is trained with L_diff = ||z - z_hat||_1 (Eq. 11) to reproduce that same residual. The 2.77 dB gain of the full model over 'w/o DM' therefore conflates the effect of the DM architecture with the two-stage joint-training protocol; the 'w/o JT' row already shows a 0.66 dB gain from joint training alone. Please clarify what the DM contributes beyond a learned regressor from the hazy amplitude spectrum to the ARE-defined target, and provide an ablation in which the DM is replaced by a deterministic network of comparable capacity trained with the same loss and protocol.
  3. [Sec. 3.3, Eq. (11)] The diffusion loss is a direct L1 loss on the final reverse sample z_hat, rather than the standard variational noise-prediction objective used in DDPMs (Ho et al., Eq. 14). With T=8 and a 0.07M-parameter denoising network, it is not clear that the model learns the intermediate noise-conditioned distributions; it may instead learn an amortized mapping from Gaussian noise to the ARE residual. Please specify the noise schedule beta_t, give training or validation curves, and either justify Eq. (11) as a valid diffusion objective or describe the method as an iterative refinement rather than a diffusion model.
  4. [Sec. 3.2, Eq. (3); Supp. A7] The ARE defines the diffusion target z by a global mean/std alignment of amplitude spectra (Eq. 3). As the authors acknowledge in Supp. A7, this ignores local spatial variations in haze thickness. Because the diffusion target is derived entirely from this global alignment, any failure of the global-statistics assumption will propagate to the DM output regardless of the dehazing network's capacity. Please provide a quantitative evaluation on scenes with strongly non-uniform haze (for example, dense local fog regions) and discuss how the network's spatial processing compensates for the mis-specified target.
minor comments (6)
  1. [Sec. 1, Contribution 1] The abstract and contribution list claim 'the first work to apply the diffusion model to the unpaired image dehazing task,' but Sec. 2.2 cites DiffIR and HiDiff as DMs that reconstruct priors or latents for restoration. Please clarify the precise difference from these methods and restrict the novelty claim accordingly.
  2. [Sec. 4.4, Table 3] The description of the 'w/o DM' baseline is ambiguous: the sentence 'When we obtained z directly using the hazy image as the ARE's input' conflicts with Eq. (3), where ARE uses both hazy and clear amplitude spectra. Please specify the exact input configuration of the w/o DM baseline.
  3. [Sec. 4.1] Table 2 reports FADE and BRISQUE without error bars; these no-reference metrics are known to be noisy, and the text describes FrDiff as 'favourable' on URHI while it is not the best. Please add standard deviations over multiple runs or bootstrap estimates and state explicitly whether lower values are better for both metrics.
  4. [Supp. A8, Fig. A4] The L2 similarity between frequency features and GT is reported as ODCR:6.257 and FrDiff:2.156, but the feature definition, the normalization, and the frequency range used for the comparison are not specified; please provide the exact computation.
  5. [Sec. 4.2] The training details specify an input patch size of 256x256 but do not state whether the SOTS-Indoor and SOTS-Outdoor evaluations are performed on full-resolution images or on cropped patches; please clarify the evaluation protocol to match the FLOPs computation in Table 1.
  6. [References] The reference list contains malformed trailing page-number fields after some entries (for example, [1] ends with '5, 6, 3'), which appear to be citation locations leaked into the bibliography; please clean up the formatting.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the DM learns to reproduce ARE-derived residuals from hazy amplitude alone and is evaluated on external benchmarks.

full rationale

FrDiff's derivation chain is self-contained in the relevant sense. The diffusion target z is defined by the deterministic ARE formula (Eq. 3) on unpaired training pairs, and the DM is trained with L_diff = ||z - zhat||_1 (Eq. 11), but at inference zhat is generated from Gaussian noise and the hazy amplitude Ah alone, without access to any clear image or to ARE. The learned component therefore must generalize to estimate the residual from the hazy amplitude, and the method is benchmarked against external SOTA numbers on held-out SOTS/HSTS/I-HAZE/O-HAZE sets. That the target is produced by the authors' own non-learned encoder is a training-signal design choice, not a case where the prediction is identical to the input by construction; a supervised regressor trained on pseudo-targets is ordinary learning. No load-bearing argument depends on a self-citation or an imported uniqueness theorem; the frequency-domain premise is supported by external references [39, 48] and the paper's own dark-channel statistics in Supp. A1. The large discrepancy between Tab. 1 (27.43 dB) and Supp. Tab. A2 (36.54 dB) for the same default configuration is a reproducibility/correctness concern, not a circularity, and does not change the circularity score.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central method rests on a small number of hand-chosen loss weights and diffusion steps, an undisclosed noise schedule, and three domain assumptions about the frequency-domain structure of haze. No novel physical entity is introduced; the amplitude residual z is a mathematical feature constructed from existing FFT statistics.

free parameters (3)
  • loss weights lambda_GAN, lambda_PatchNCE, lambda_diff = all set to 1
    Chosen by hand in Sec. 4.2; supplementary Tab. A2 shows performance varies with these weights, so the reported result depends on this choice.
  • diffusion steps T = 8
    Selected by ablation in Tab. 7 as a trade-off between PSNR and FLOPs; performance increases monotonically with T.
  • diffusion noise schedule beta_t = not specified
    The paper says beta_t are hyperparameters but never gives the actual schedule (Sec. 3.3, Eq. 7); the result depends on an undisclosed schedule.
assumptions (4)
  • domain assumption Haze degradation is mainly manifested in the amplitude spectrum, while phase preserves texture structure.
    Used to justify reconstructing only amplitude residuals; supported only by qualitative dark-channel histograms in Supp. A1, not by a quantitative proof.
  • ad hoc to paper Global mean/std alignment of amplitude spectra (Eq. 3) is sufficient to bridge hazy and clear amplitude distributions.
    A'_h is a linear rescaling of A_h using per-sample statistics; there is no argument that first-order statistics capture all haze-relevant amplitude information.
  • ad hoc to paper Direct L1 supervision on the final reverse diffusion sample (Eq. 11) is a valid training objective for the denoising network.
    Standard DDPM training uses a noise-prediction loss at random timesteps; this paper instead backpropagates through the full reverse chain, which is not justified.
  • domain assumption The synthetic ITS training set generalizes to real-world hazy images.
    Evaluation on HSTS-Real, Fattal, and URHI uses no-reference metrics; domain shift from synthetic training is assumed manageable.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Frequency Domain-Based Diffusion Model for Unpaired Image Dehazing." pith.science (2026). https://pith.science/paper/7KDPMADD

@misc{pith2026250701275,
  author       = {Pith},
  title        = {Pith review of: Frequency Domain-Based Diffusion Model for Unpaired Image Dehazing},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7KDPMADD}},
  note         = {Machine review of arXiv:2507.01275}
}
read the original abstract

Unpaired image dehazing has attracted increasing attention due to its flexible data requirements during model training. Dominant methods based on contrastive learning not only introduce haze-unrelated content information, but also ignore haze-specific properties in the frequency domain (\ie,~haze-related degradation is mainly manifested in the amplitude spectrum). To address these issues, we propose a novel frequency domain-based diffusion model, named \ours, for fully exploiting the beneficial knowledge in unpaired clear data. In particular, inspired by the strong generative ability shown by Diffusion Models (DMs), we tackle the dehazing task from the perspective of frequency domain reconstruction and perform the DMs to yield the amplitude spectrum consistent with the distribution of clear images. To implement it, we propose an Amplitude Residual Encoder (ARE) to extract the amplitude residuals, which effectively compensates for the amplitude gap from the hazy to clear domains, as well as provide supervision for the DMs training. In addition, we propose a Phase Correction Module (PCM) to eliminate artifacts by further refining the phase spectrum during dehazing with a simple attention mechanism. Experimental results demonstrate that our \ours outperforms other state-of-the-art methods on both synthetic and real-world datasets.

Figures

Figures reproduced from arXiv: 2507.01275 by the authors.

Figure 1
Figure 1. Comparison of performance, FLOPs, and parameters [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. (a) Illustration of the property of haze degradation in the frequency domain. Haze degradation can be transferred with the [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Schematic of training and inference strategies. [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: (a) Overview of FrDiff, which mainly consists of a dehazing network and a diffusion model (DM). (b) Structure of the Amplitude [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 5
Figure 5. Figure 5: Visual results on SOTS-Indoor [28] and SOTS-Outdoor [28] datasets. Zoom in to see better visualization. Hazy RefineDNet CDD-GAN UCL-Dehaze ODCR Ours Fattal's HSTS -Real URHI D 4 [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Visual results on real-world HSTS-Real [ [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: Visualization of ablation study on FrDiff. [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

71 extracted references · 62 canonical work pages

  1. [1]

    I-HAZE: A dehazing bench- mark with real hazy and haze-free indoor images

    Cosmin Ancuti, Codruta O Ancuti, Radu Timofte, and Christophe De Vleeschouwer. I-HAZE: A dehazing bench- mark with real hazy and haze-free indoor images. In ACIVS, pages 620–631. Springer, 2018. 5, 6, 3

    work page 2018

  2. [2]

    O-HAZE: a dehazing bench- mark with real hazy and haze-free outdoor images

    Codruta O Ancuti, Cosmin Ancuti, Radu Timofte, and Christophe De Vleeschouwer. O-HAZE: a dehazing bench- mark with real hazy and haze-free outdoor images. In CVPRW, pages 754–762, 2018. 4

    work page 2018

  3. [3]

    Non-local image dehazing

    Dana Berman, Shai Avidan, et al. Non-local image dehazing. In CVPR, pages 1674–1682, 2016. 1

    work page 2016

  4. [4]

    Single im- age dehazing using haze-lines.IEEE TPAMI, 42(3):720–734,

    Dana Berman, Tali Treibitz, and Shai Avidan. Single im- age dehazing using haze-lines.IEEE TPAMI, 42(3):720–734,

    work page

  5. [5]

    DehazeNet: An end-to-end system for single image haze removal

    Bolun Cai, Xiangmin Xu, Kui Jia, Chunmei Qing, and Dacheng Tao. DehazeNet: An end-to-end system for single image haze removal. IEEE TIP, 25(11):5187–5198, 2016. 2

    work page 2016

  6. [6]

    Simple baselines for image restoration

    Liangyu Chen, Xiaojie Chu, Xiangyu Zhang, and Jian Sun. Simple baselines for image restoration. In ECCV, pages 17–

  7. [7]

    Unpaired deep image dehazing using contrastive disentan- glement learning

    Xiang Chen, Zhentao Fan, Pengpeng Li, Longgang Dai, Cai- hua Kong, Zhuoran Zheng, Yufeng Huang, and Yufeng Li. Unpaired deep image dehazing using contrastive disentan- glement learning. In ECCV, pages 632–648. Springer, 2022. 1, 3, 6

    work page 2022

  8. [8]

    PSD: Principled synthetic-to-real dehazing guided by physi- cal priors

    Zeyuan Chen, Yangchao Wang, Yang Yang, and Dong Liu. PSD: Principled synthetic-to-real dehazing guided by physi- cal priors. In CVPR, pages 7180–7189, 2021. 3, 6

    work page 2021

Show all 71 references
  1. [9]

    Prompt-based test-time real image dehazing: a novel pipeline

    Zixuan Chen, Zewei He, Ziqian Lu, Xuecheng Sun, and Zhe- Ming Lu. Prompt-based test-time real image dehazing: a novel pipeline. In ECCV, pages 432–449. Springer, 2024. 3, 6

    2024

  2. [10]

    Hierarchical integration diffusion model for realistic image deblurring

    Zheng Chen, Yulun Zhang, Ding Liu, Jinjin Gu, Linghe Kong, Xin Yuan, et al. Hierarchical integration diffusion model for realistic image deblurring. NeurIPS, 36, 2024. 2, 3, 5

    2024

  3. [11]

    Ref- erenceless prediction of perceptual fog density and percep- tual image defogging

    Lark Kwon Choi, Jaehee You, and Alan Conrad Bovik. Ref- erenceless prediction of perceptual fog density and percep- tual image defogging. IEEE TIP, 24(11):3888–3901, 2015. 7

    2015

  4. [12]

    HardGAN: A haze-aware representation distillation gan for single image dehazing

    Qili Deng, Ziling Huang, Chung-Chi Tsai, and Chia-Wen Lin. HardGAN: A haze-aware representation distillation gan for single image dehazing. In ECCV, pages 722–738. Springer, 2020. 2

    2020

  5. [13]

    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. In CVPR, pages 2157–2167, 2020. 1, 2

    2020

  6. [14]

    An image is worth 16x16 words: Trans- formers for image recognition at scale

    Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Syl- vain Gelly, et al. An image is worth 16x16 words: Trans- formers for image recognition at scale. In ICLR, 2020. 8

    2020

  7. [15]

    Cycle- Dehaze: Enhanced cyclegan for single image dehazing

    Deniz Engin, Anil Genc ¸, and Hazim Kemal Ekenel. Cycle- Dehaze: Enhanced cyclegan for single image dehazing. In CVPRW, pages 825–833, 2018. 1, 3

    2018

  8. [16]

    Neural sparse representa- tion for image restoration

    Yuchen Fan, Jiahui Yu, Yiqun Mei, Yulun Zhang, Yun Fu, Ding Liu, and Thomas S Huang. Neural sparse representa- tion for image restoration. NeurIPS, 33:15394–15404, 2020. 8

    2020

  9. [17]

    Real-world image dehazing with coherence-based pseudo la- beling and cooperative unfolding network

    Chengyu Fang, Chunming He, Fengyang Xiao, Yulun Zhang, Longxiang Tang, Yuelin Zhang, Kai Li, and Xiu Li. Real-world image dehazing with coherence-based pseudo la- beling and cooperative unfolding network. NeurIPS, 37: 97859–97883, 2025. 3, 6

    2025

  10. [18]

    Dehazing using color-lines

    Raanan Fattal. Dehazing using color-lines. ACM TOG, 34 (1):1–14, 2014. 1, 2, 5, 6, 7, 3, 10

    2014

  11. [19]

    Single image haze removal using dark channel prior

    Kaiming He, Jian Sun, and Xiaoou Tang. Single image haze removal using dark channel prior. IEEE TPAMI, 33(12): 2341–2353, 2010. 1, 2, 6

    2010

  12. [20]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In CVPR, pages 770–778, 2016. 5

    2016

  13. [21]

    Denoising diffu- sion probabilistic models

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising diffu- sion probabilistic models. NeurIPS, 33:6840–6851, 2020. 3, 5

    2020

  14. [22]

    Squeeze-and-excitation net- works

    Jie Hu, Li Shen, and Gang Sun. Squeeze-and-excitation net- works. In CVPR, pages 7132–7141, 2018. 8

    2018

  15. [23]

    Spatial transformer networks

    Max Jaderberg, Karen Simonyan, Andrew Zisserman, et al. Spatial transformer networks. NeurIPS, 28, 2015. 8

    2015

  16. [24]

    Denoising diffusion restoration models

    Bahjat Kawar, Michael Elad, Stefano Ermon, and Jiaming Song. Denoising diffusion restoration models. NeurIPS, 35: 23593–23606, 2022. 2, 3

    2022

  17. [25]

    Auto-encoding variational bayes

    Diederik P Kingma. Auto-encoding variational bayes. In ICLR, 2014. 5

    2014

  18. [26]

    Adam: A method for stochastic optimization

    Diederik P Kingma and Jimmy Ba. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980 ,

    arXiv

  19. [27]

    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. In ICCV, pages 4770–4778, 2017. 2, 6

    2017

  20. [28]

    Benchmarking single- image dehazing and beyond

    Boyi Li, Wenqi Ren, Dengpan Fu, Dacheng Tao, Dan Feng, Wenjun Zeng, and Zhangyang Wang. Benchmarking single- image dehazing and beyond. IEEE TIP , 28(1):492–505,

  21. [29]

    You only look yourself: Un- supervised and untrained single image dehazing neural net- work

    Boyun Li, Yuanbiao Gou, Shuhang Gu, Jerry Zitao Liu, Joey Tianyi Zhou, and Xi Peng. You only look yourself: Un- supervised and untrained single image dehazing neural net- work. IJCV, 129:1754–1767, 2021. 6

    2021

  22. [30]

    Embedding fourier for ultra-high-definition low-light image enhancement

    Chongyi Li, Chun-Le Guo, Man Zhou, Zhexin Liang, Shangchen Zhou, Ruicheng Feng, and Chen Change Loy. Embedding fourier for ultra-high-definition low-light image enhancement. In ICLR, 2023. 1

    2023

  23. [31]

    Detection- friendly dehazing: Object detection in real-world hazy scenes

    Chengyang Li, Heng Zhou, Yang Liu, Caidong Yang, Yongqiang Xie, Zhongbo Li, and Liping Zhu. Detection- friendly dehazing: Object detection in real-world hazy scenes. IEEE TPAMI, 45(7):8284–8295, 2023. 1

    2023

  24. [32]

    Rethinking diffusion model for multi-contrast mri super-resolution

    Guangyuan Li, Chen Rao, Juncheng Mo, Zhanjie Zhang, Wei Xing, and Lei Zhao. Rethinking diffusion model for multi-contrast mri super-resolution. In CVPR, pages 11365– 11374, 2024. 2, 3, 5

    2024

  25. [33]

    Springer, 2022. 3, 2

    2022

  26. [34]

    Grid- dehazenet: Attention-based multi-scale network for image dehazing

    Xiaohong Liu, Yongrui Ma, Zhihao Shi, and Jun Chen. Grid- dehazenet: Attention-based multi-scale network for image dehazing. In ICCV, pages 7314–7323, 2019. 1, 2

    2019

  27. [35]

    End- to-end single image fog removal using enhanced cycle con- 11 sistent adversarial networks

    Wei Liu, Xianxu Hou, Jiang Duan, and Guoping Qiu. End- to-end single image fog removal using enhanced cycle con- 11 sistent adversarial networks. IEEE TIP , 29:7819–7833,

  28. [36]

    Repaint: Inpainting using denoising diffusion probabilistic models

    Andreas Lugmayr, Martin Danelljan, Andres Romero, Fisher Yu, Radu Timofte, and Luc Van Gool. Repaint: Inpainting using denoising diffusion probabilistic models. In CVPR, pages 11461–11471, 2022. 2, 3

    2022

  29. [37]

    Learning deep priors for image dehazing

    Yang Liu, Jinshan Pan, Jimmy Ren, and Zhixun Su. Learning deep priors for image dehazing. In ICCV, pages 2492–2500,

  30. [38]

    Vision and the atmosphere

    Srinivasa G Narasimhan and Shree K Nayar. Vision and the atmosphere. IJCV, 48:233–254, 2002. 1

    2002

  31. [39]

    No-reference image quality assessment in the spa- tial domain

    Anish Mittal, Anush Krishna Moorthy, and Alan Conrad Bovik. No-reference image quality assessment in the spa- tial domain. IEEE TIP, 21(12):4695–4708, 2012. 7

    2012

  32. [40]

    Contrastive learning for unpaired image-to-image translation

    Taesung Park, Alexei A Efros, Richard Zhang, and Jun- Yan Zhu. Contrastive learning for unpaired image-to-image translation. In ECCV, pages 319–345. Springer, 2020. 1, 3, 5, 6, 4

    2020

  33. [41]

    The importance of phase in signals

    Alan V Oppenheim and Jae S Lim. The importance of phase in signals. Proceedings of the IEEE , 69(5):529–541, 1981. 1, 2, 4

    1981

  34. [42]

    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. In AAAI, pages 11908–11915, 2020. 1, 2, 6

    2020

  35. [43]

    Learning depth-density priors for fourier-based unpaired image restoration

    Yuanjian Qiao, Mingwen Shao, Leiquan Wang, and Wang- meng Zuo. Learning depth-density priors for fourier-based unpaired image restoration. IEEE TCSVT, 2023. 1, 3

    2023

  36. [44]

    Single image dehazing via multi- scale convolutional neural networks

    Wenqi Ren, Si Liu, Hua Zhang, Jinshan Pan, Xiaochun Cao, and Ming-Hsuan Yang. Single image dehazing via multi- scale convolutional neural networks. In ECCV, pages 154–

  37. [45]

    En- hanced pix2pix dehazing network

    Yanyun Qu, Yizi Chen, Jingying Huang, and Yuan Xie. En- hanced pix2pix dehazing network. In CVPR, pages 8160– 8168, 2019. 6

    2019

  38. [46]

    Model adaptation with synthetic and real data for semantic dense foggy scene understanding

    Christos Sakaridis, Dengxin Dai, Simon Hecker, and Luc Van Gool. Model adaptation with synthetic and real data for semantic dense foggy scene understanding. In ECCV, pages 687–704, 2018. 1

    2018

  39. [47]

    Domain adaptation for image dehazing

    Yuanjie Shao, Lerenhan Li, Wenqi Ren, Changxin Gao, and Nong Sang. Domain adaptation for image dehazing. In CVPR, pages 2808–2817, 2020. 2, 3

    2020

  40. [48]

    Gated fusion net- work for single image dehazing

    Wenqi Ren, Lin Ma, Jiawei Zhang, Jinshan Pan, Xiaochun Cao, Wei Liu, and Ming-Hsuan Yang. Gated fusion net- work for single image dehazing. InCVPR, pages 3253–3261,

  41. [49]

    Denois- ing diffusion implicit models

    Jiaming Song, Chenlin Meng, and Stefano Ermon. Denois- ing diffusion implicit models. In ICLR, 2021. 5

    2021

  42. [50]

    Vision transformers for single image dehazing.IEEE TIP, 32:1927– 1941, 2023

    Yuda Song, Zhuqing He, Hui Qian, and Xin Du. Vision transformers for single image dehazing.IEEE TIP, 32:1927– 1941, 2023. 2

    1927

  43. [51]

    The importance of phase in image processing

    Nikolay Skarbnik, Yehoshua Y Zeevi, and Chen Sagiv. The importance of phase in image processing . Technion-Israel Institute of Technology, Faculty of Electrical Engineering,

  44. [52]

    Attention is all you need

    A Vaswani. Attention is all you need. NeurIPS, 2017. 5, 8

    2017

  45. [53]

    Ucl-dehaze: Towards real-world image dehazing via unsupervised contrastive learning

    Yongzhen Wang, Xuefeng Yan, Fu Lee Wang, Haoran Xie, Wenhan Yang, Xiao-Ping Zhang, Jing Qin, and Mingqiang Wei. Ucl-dehaze: Towards real-world image dehazing via unsupervised contrastive learning. IEEE TIP, 2024. 1, 3, 5, 6, 4

    2024

  46. [54]

    Neural discrete representation learning

    Aaron Van Den Oord, Oriol Vinyals, et al. Neural discrete representation learning. NeurIPS, 30, 2017. 8

    2017

  47. [55]

    Cbam: Convolutional block attention module

    Sanghyun Woo, Jongchan Park, Joon-Young Lee, and In So Kweon. Cbam: Convolutional block attention module. In ECCV, pages 3–19, 2018. 8

    2018

  48. [56]

    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. In CVPR, pages 10551–10560, 2021. 2

    2021

  49. [57]

    Odcr: Orthogonal decoupling contrastive regularization for unpaired image dehazing

    Zhongze Wang, Haitao Zhao, Jingchao Peng, Lujian Yao, and Kaijie Zhao. Odcr: Orthogonal decoupling contrastive regularization for unpaired image dehazing. In CVPR, pages 25479–25489, 2024. 1, 3, 5, 6, 4

    2024

  50. [58]

    Unsupervised feature learning via non-parametric instance discrimination

    Zhirong Wu, Yuanjun Xiong, Stella X Yu, and Dahua Lin. Unsupervised feature learning via non-parametric instance discrimination. In CVPR, pages 3733–3742, 2018. 8

    2018

  51. [59]

    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 ICCV, pages 13095–13105, 2023. 2, 3, 5

    2023

  52. [60]

    RIDCP: Revitalizing real image dehazing via high-quality codebook priors

    Rui-Qi Wu, Zheng-Peng Duan, Chun-Le Guo, Zhi Chai, and Chongyi Li. RIDCP: Revitalizing real image dehazing via high-quality codebook priors. In CVPR, pages 22282– 22291, 2023. 2, 3, 6

    2023

  53. [61]

    Learning diffusion texture priors for image restoration

    Tian Ye, Sixiang Chen, Wenhao Chai, Zhaohu Xing, Jing Qin, Ge Lin, and Lei Zhu. Learning diffusion texture priors for image restoration. In CVPR, pages 2524–2534, 2024. 2, 3

    2024

  54. [62]

    Frequency and spatial dual guidance for image dehazing

    Hu Yu, Naishan Zheng, Man Zhou, Jie Huang, Zeyu Xiao, and Feng Zhao. Frequency and spatial dual guidance for image dehazing. In ECCV, pages 181–198. Springer, 2022. 1, 3

    2022

  55. [63]

    Self-augmented unpaired image dehazing via density and depth decomposition

    Yang Yang, Chaoyue Wang, Risheng Liu, Lin Zhang, Xiao- jie Guo, and Dacheng Tao. Self-augmented unpaired image dehazing via density and depth decomposition. In CVPR, pages 2037–2046, 2022. 1, 3, 6, 4

    2022

  56. [64]

    Densely connected pyramid dehazing network

    He Zhang and Vishal M Patel. Densely connected pyramid dehazing network. In CVPR, pages 3194–3203, 2018. 1, 2

    2018

  57. [65]

    Re- finednet: A weakly supervised refinement framework for sin- gle image dehazing

    Shiyu Zhao, Lin Zhang, Ying Shen, and Yicong Zhou. Re- finednet: A weakly supervised refinement framework for sin- gle image dehazing. IEEE TIP, 30:3391–3404, 2021. 6

    2021

  58. [66]

    High- quality image dehazing with diffusion model

    Hu Yu, Jie Huang, Kaiwen Zheng, and Feng Zhao. High- quality image dehazing with diffusion model. arXiv preprint arXiv:2308.11949, 2023. 2

    2023 arXiv

  59. [67]

    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. In ICCV, pages 2223–2232,

  60. [68]

    A fast single image haze removal algorithm using color attenuation prior

    Qingsong Zhu, Jiaming Mai, and Ling Shao. A fast single image haze removal algorithm using color attenuation prior. IEEE TIP, 24(11):3522–3533, 2015. 1, 2 13

    2015

  61. [69]

    Curricular contrastive regularization for physics- aware single image dehazing

    Yu Zheng, Jiahui Zhan, Shengfeng He, Junyu Dong, and Yong Du. Curricular contrastive regularization for physics- aware single image dehazing. In CVPR, pages 5785–5794,

  62. [169]

    Springer, 2016. 1, 2, 6

    2016

  63. [2018]

    1, 5, 6, 7, 8, 3, 4, 9, 10

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.