Denoising Diffusion Models for Plug-and-Play Image Restoration
Open accessThe DOI matches its Crossref record; 33 of the 33 references it cites were found in a registry or an index.
Abstract
Plug-and-play Image Restoration (IR) has been widely recognized as a flexible and interpretable method for solving various inverse problems by utilizing any off-the-shelf denoiser as the implicit image prior. However, most existing methods focus on discriminative Gaussian denoisers. Although diffusion models have shown impressive performance for high-quality image synthesis, their potential to serve as a generative denoiser prior to the plug-and-play IR methods remains to be further explored. While several other attempts have been made to adopt diffusion models for image restoration, they either fail to achieve satisfactory results or typically require an unacceptable number of Neural Function Evaluations (NFEs) during inference. This paper proposes DiffPIR, which integrates the traditional plug-and-play method into the diffusion sampling framework. Compared to plug-and-play IR methods that rely on discriminative Gaussian denoisers, DiffPIR is expected to inherit the generative ability of diffusion models. Experimental results on three representative IR tasks, including super-resolution, image deblurring, and inpainting, demonstrate that DiffPIR achieves state-of-the-art performance on both the FFHQ and ImageNet datasets in terms of reconstruction faithfulness and perceptual quality with no more than 100 NFEs. The source code is available at https://github.com/yuanzhi-zhu/DiffPIR
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Auto-Encoding Variational Bayes
Kingma, D. P., & Welling, M. (2013) · arXiv · doi:10.48550/arxiv.1312.6114 · Cited by 20,067
Generative adversarial networks
Goodfellow, I., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville, A., & Bengio, Y. (2020) · Communications of the ACM · doi:10.1145/3422622 · Cited by 18,384
Boyd, S., Parikh, N., Chu, E., Peleato, B., & Eckstein, J. (2011) · Foundations and Trends® in Machine Learning · doi:10.1561/2200000016 · Cited by 17,075
Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network
Ledig, C., Theis, L., Huszar, F., Caballero, J., Cunningham, A., Acosta, A., Aitken, A., Tejani, A., Totz, J., Wang, Z., et al. (2017) · 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2017.19 · Cited by 12,696
A Style-Based Generator Architecture for Generative Adversarial Networks
Karras, T., Laine, S., & Aila, T. (2019) · 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2019.00453 · Cited by 11,714 · cited 8×
“Deep generative models such as Generative Adversarial Networks (GANs) [23,31], Normalizing Flows (NFs) [15] and Variational Autoencoders (VAEs) [33, 44] have been used as denoisers of plug-and-play IR framework [18, 35, 55].”
MethodInfluentialImage Super-Resolution Using Deep Convolutional Networks
Dong, C., Loy, C. C., He, K., & Tang, X. (2016) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2015.2439281 · Cited by 10,105
Denoising Diffusion Probabilistic Models
Ho, J., Jain, A., & Abbeel, P. (2020) · arXiv · doi:10.48550/arxiv.2006.11239 · Cited by 8,712
A Non-Local Algorithm for Image Denoising
Buades, A., Coll, B., & Morel, J.-M. (2005) · 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05) · doi:10.1109/cvpr.2005.38 · Cited by 7,616 · cited 1×
“Recent studies have demonstrated that plug-and-play Image Restoration (IR) methods can effectively handle a variety of low-level vision tasks, such as image denoising [5], image Super-Resolution (SR) [16, 17, 37], image deblurring [12] and image inpainting [27], with excellent results [6, 11, 54, 57, 58].”
BackgroundParikh, N., & Boyd, S. (2014) · Foundations and Trends® in Optimization · doi:10.1561/2400000003 · Cited by 4,102
Diffusion Models Beat GANs on Image Synthesis
Dhariwal, P., & Nichol, A. (2021) · arXiv · doi:10.48550/arxiv.2105.05233 · Cited by 3,362
Deep Unsupervised Learning using Nonequilibrium Thermodynamics
Sohl-Dickstein, J., Weiss, E. A., Maheswaranathan, N., & Ganguli, S. (2015) · arXiv · doi:10.48550/arxiv.1503.03585 · Cited by 2,637
Hierarchical Text-Conditional Image Generation with CLIP Latents
Ramesh, A., Dhariwal, P., Nichol, A., Chu, C., & Chen, M. (2022) · arXiv · doi:10.48550/arxiv.2204.06125 · Cited by 2,511 · cited 1×
“Recently, diffusion models have demonstrated the ability to generate images with higher quality [14, 41, 43] than previous generative models such as GANs, VAEs and NFs even when starting from pure Gaussian noise.”
BackgroundDenoising Diffusion Implicit Models
Song, J., Meng, C., & Ermon, S. (2020) · arXiv · doi:10.48550/arxiv.2010.02502 · Cited by 2,507 · cited 6×
“To be specific, we adapt the quadratic sequence in DDIM for sampling, which has more sampling steps at low-noise regions [51].”
BackgroundMethodInfluentialGlobally and locally consistent image completion
Iizuka, S., Simo-Serra, E., & Ishikawa, H. (2017) · ACM Transactions on Graphics · doi:10.1145/3072959.3073659 · Cited by 2,203 · cited 1×
“…have demonstrated that plug-and-play Image Restoration (IR) methods can effectively handle a variety of low-level vision tasks, such as image denoising [5,58], image Super-Resolution (SR) [16,17,37], image deblurring [12] and image inpainting [27], with excellent results [6, 11, 54, 57, 59].”
BackgroundLearning Deep CNN Denoiser Prior for Image Restoration
Zhang, K., Zuo, W., Gu, S., & Zhang, L. (2017) · 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2017.300 · Cited by 2,116 · cited 2×
“[58] made the first attempt to incorporate deep denoiser priors to solve various IR tasks.”
BackgroundSelf-citationScore-Based Generative Modeling through Stochastic Differential Equations
Song, Y., Sohl-Dickstein, J., Kingma, D. P., Kumar, A., Ermon, S., & Poole, B. (2020) · arXiv · doi:10.48550/arxiv.2011.13456 · Cited by 1,953 · cited 8×
“In this way, the total variance is preserved and DDPM is also called ”Variance Preserving” (VP) SDE [53].”
BackgroundInfluentialImage Super-Resolution Via Iterative Refinement
Saharia, C., Ho, J., Chan, W., Salimans, T., Fleet, D. J., & Norouzi, M. (2022) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2022.3204461 · Cited by 1,829 · cited 1×
“[48, 49] trained image-conditional diffusion models for SR and image-to-image translation in concurrent work.”
Image Super-Resolution via Iterative Refinement
Saharia, C., Ho, J., Chan, W., Salimans, T., Fleet, D. J., & Norouzi, M. (2021) · arXiv · doi:10.48550/arxiv.2104.07636 · Cited by 1,829
RePaint: Inpainting using Denoising Diffusion Probabilistic Models
Lugmayr, A., Danelljan, M., Romero, A., Yu, F., Timofte, R., & Van Gool, L. (2022) · 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr52688.2022.01117 · Cited by 1,778 · cited 5×
“, [7,8,38] and sampling methods in Appendix A.”
BackgroundMethodInfluentialSelf-citationPalette: Image-to-Image Diffusion Models
Saharia, C., Chan, W., Chang, H., Lee, C., Ho, J., Salimans, T., Fleet, D., & Norouzi, M. (2022) · Special Interest Group on Computer Graphics and Interactive Techniques Conference Proceedings · doi:10.1145/3528233.3530757 · Cited by 1,717
Improved Denoising Diffusion Probabilistic Models
Nichol, A., & Dhariwal, P. (2021) · arXiv · doi:10.48550/arxiv.2102.09672 · Cited by 1,550 · cited 1×
“Recently, diffusion models have demonstrated the ability to generate images with higher quality [14, 41, 43] than previous generative models such as GANs, VAEs and NFs even when starting from pure Gaussian noise.”
BackgroundGenerative Modeling by Estimating Gradients of the Data Distribution
Song, Y., & Ermon, S. (2019) · arXiv · doi:10.48550/arxiv.1907.05600 · Cited by 1,453 · cited 1×
“…in (2) can be reversed in time and has the form of [2]: where p t ( x ) is the marginal probability density at timestep t , and the only unknown part ∇ x log p t ( x ) can be modelled as so-called score function s θ ( x , t ) with score matching methods [26, 52]. x at t is denoted as x t .”
MethodGLIDE: Towards Photorealistic Image Generation and Editing with Text-Guided Diffusion Models
Nichol, A., Dhariwal, P., Ramesh, A., Shyam, P., Mishkin, P., McGrew, B., Sutskever, I., & Chen, M. (2021) · arXiv · doi:10.48550/arxiv.2112.10741 · Cited by 1,357 · cited 1×
“[40] propose to use text-guided diffusion models to generate photo-realistic images with classifier-free guidance.”
MethodPlug-and-Play priors for model based reconstruction
Venkatakrishnan, S. V., Bouman, C. A., & Wohlberg, B. (2013) · 2013 IEEE Global Conference on Signal and Information Processing · doi:10.1109/globalsip.2013.6737048 · Cited by 1,178 · cited 3×
“Recent studies have demonstrated that plug-and-play Image Restoration (IR) methods can effectively handle a variety of low-level vision tasks, such as image denoising [5], image Super-Resolution (SR) [16, 17, 37], image deblurring [12] and image inpainting [27], with excellent results [6, 11, 54, 57, 58].”
Classifier-Free Diffusion Guidance
Ho, J., & Salimans, T. (2022) · arXiv · doi:10.48550/arxiv.2207.12598 · Cited by 1,126 · cited 1×
“[25] introduced the classifier-free diffusion guidance with s e (x ,f, y) = V x logp t (x |y) the image conditional diffusion models.”
MethodNonlinear image recovery with half-quadratic regularization
Geman, D., & Yang, C. (1995) · IEEE Transactions on Image Processing · doi:10.1109/83.392335 · Cited by 1,000 · cited 1×
“With the help of variable splitting algorithms, such as the Alternating Direction Method of Multipliers (ADMM) [4] and Half-Quadratic-Splitting (HQS) [22], plug-and-play IR methods integrate Gaussian denoisers into the iterative process, leading to improved performance and convergence.”
MethodThe Little Engine That Could: Regularization by Denoising (RED)
Romano, Y., Elad, M., & Milanfar, P. (2017) · SIAM Journal on Imaging Sciences · doi:10.1137/16m1102884 · Cited by 934 · cited 1×
“[19] proposed a Moreau-Yosida regularised unadjusted Langevin algorithm for Bayesian computation such as imaging inverse problems.”
MethodPlug-and-Play ADMM for Image Restoration: Fixed-Point Convergence and Applications
Chan, S. H., Wang, X., & Elgendy, O. A. (2017) · IEEE Transactions on Computational Imaging · doi:10.1109/tci.2016.2629286 · Cited by 904
(2005) · doi:10.5555/1046920.1088696 · Cited by 756
Generating diverse high-fidelity images with vq-vae-2
Razavi, A., Oord, A. V. D., & Vinyals, O. (2019) · arXiv · doi:10.48550/arxiv.1906.00446 · Cited by 709
Elucidating the Design Space of Diffusion-Based Generative Models
Karras, T., Aittala, M., Aila, T., & Laine, S. (2022) · arXiv · doi:10.48550/arxiv.2206.00364 · Cited by 702
BM3D Frames and Variational Image Deblurring
Danielyan, A., Katkovnik, V., & Egiazarian, K. (2012) · IEEE Transactions on Image Processing · doi:10.1109/tip.2011.2176954 · Cited by 611
Reverse-time diffusion equation models
Anderson, B. D. O. (1982) · Stochastic Processes and their Applications · doi:10.1016/0304-4149(82)90051-5 · Cited by 565 · cited 1×
“The diffusion process described in (2) can be reversed in time and has the form of [2]: where p t ( x ) is the marginal probability density at timestep t , and the only unknown part ∇ x log p t ( x ) can be modelled as so-called score function s θ ( x , t ) with score matching methods [26, 52]. x…”
Method
Cited by
Learning Spatial Adaptation and Temporal Coherence in Diffusion Models for Video Super-Resolution
Chen, Z., Long, F., Qiu, Z., Yao, T., Zhou, W., Luo, J., & Mei, T. (2024) · 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr52733.2024.00882 · Cited by 27
“These explorations [9, 10, 13, 21, 31, 42, 49, 57] leverage the knowledge prior embedded in the pre-trained diffusion models to magnify images.”
Background
Metadata
- Authors
- Zhu, Y., Zhang, K., Liang, J., Cao, J., Wen, B., Timofte, R., & Gool, L. V.
- Venue
- 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)
- Publisher
- IEEE
- Published
- 2023-06
- Pages
- 1219-1229
- OpenAlex
- W4385804996
- Cited by
- 228
- References
- 33
- FWCI
- 25.92
- Type
- conference paper
- Language
- en
- Open access
- green
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.