Pith. sign in

REVIEW 4 major objections 5 minor 61 references

Degradation-Aware Metric Prompting for Hyperspectral Image Restoration

T0 review · 4 major / 5 minor · reviewed 2026-08-03 · deepseek-v4-flash

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

Pith's one-line read A label-free model can unify HSI restoration by routing on six interpretable degradation statistics.

desk verdict DAMP's metric-based MoE for HSI restoration is a real and useful step; the mixed-degradation and true generalization claims need experiments before they can be believed. read the letter →

arxiv 2512.20251 v3 pith:NX2TK4AW submitted 2025-12-23 cs.CV eess.IV

Binfeng Wang , Di Wang , Haonan Guo , Ying Fu , Jing Zhang This is my paper
classification cs.CVeess.IV
keywords hyperspectralimagerestorationall-in-onedegradationpromptsmixture-of-expertsspatial-spectralmetricszero-shotgeneralizationdenoisingsuper-resolution
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 tries to establish that a single hyperspectral image restoration model can handle many degradations — noise, blur, low resolution, missing regions, missing bands, and even unseen corruptions — without being told what the degradation is. The proposed DAMP framework replaces explicit degradation type labels with six hand-crafted spatial and spectral statistics computed directly from the corrupted input. Those statistics, projected into a Degradation Prompt, act as a router that selects which specialist module in a mixture-of-experts should restore the image. On five unified restoration tasks the approach beats task-specific and prompt-based baselines on most metrics, and it transfers zero-shot to two unseen degradation types. If the claim holds, it matters because real-world HSI corruption is mixed and unknown in advance, and a label-free model could be deployed without task information.

What carries the argument

Degradation Prompts (DP): six interpretable spatial-spectral statistics computed on the corrupted input, projected to a high-dimensional embedding and used as a gating router. The router selects the top-1 expert in the Degradation-Adaptive Mixture-of-Experts (DAMoE); each expert is a Spatial-Spectral Adaptive Module (SSAM) with learnable fusion weights λ_s and λ_c that balance a spatial attention branch and a spectral 1D-convolution branch, with λ_s+λ_c=1. The combination lets the model specialize experts to degradation profiles without any label, and the DP router is what aligns an input's metric signature with the trained expert. The paper also shows the same DP router can be dropped into

What would settle it

Compute the six DP metrics for a degradation the model never trained on (e.g., stripe noise or compression artifacts, or a blur+noise mixture) and compare zero-shot restoration with and without the DP router; if the DP-routed model does not outperform the router-ablated baseline, the metric transfer premise is false. A simpler observational check: plot PSNR gain against the distance of the unseen sample's DP embedding from the five training-family clusters.

Watch

Extended reading notes

Core claim

The central claim is that measurable, interpretable statistics of a degraded hyperspectral image carry enough information to identify the degradation and choose the right restoration strategy, so explicit degradation labels are unnecessary. DAMP computes six metrics — high-frequency energy ratio, spatial texture uniformity, spectral curvature mean and standard deviation, gradient standard deviation, and spatial correlation coefficient — from the input, projects them into a Degradation Prompt, and uses that prompt as the gating signal for a Degradation-Adaptive Mixture-of-Experts. The experts are Spatial-Spectral Adaptive Modules that learn separate fusion coefficients for spatial and spectra

Load-bearing premise

The load-bearing premise is that six hand-picked statistics, selected on a labeled subset of five degradation families, keep separating unseen or mixed degradation mechanisms well enough that the router lands them on the right expert; if an unseen corruption's metric signature falls outside the trained range, the router collapses to a near-random expert choice and the unified advantage disappears.

Editorial extensions

If this is right

  • A label-free unified HSI restorer reaches or beats task-specific specialists and prior prompt-based methods on Gaussian denoising, Gaussian deblurring, super-resolution, inpainting, and band completion.
  • The model generalizes zero-shot to unseen degradation types: motion deblurring and Poisson denoising improve over the best prior unified model by 0.52 dB and 2.10 dB PSNR respectively.
  • Degradation Prompts are plug-and-play: wiring the same DP router into two existing unified architectures raises their average PSNR by 1.00 dB and 1.99 dB.
  • The full framework adds only about 0.15 GFLOPs of overhead over its own base model, so the degradation-awareness comes nearly for free.
  • Multi-task training on all five degradations improves four of five individual tasks over single-task training, suggesting shared degradation statistics help rather than hurt.

Reading between the lines

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

  • If the metric manifold generalizes beyond the five training families, the same six statistics could serve as a universal degradation descriptor for other signal-restoration domains — for example, using DP-style routing in RGB restoration or medical imaging — but that is an extrapolation beyond the paper's evidence.
  • The strongest test the paper leaves open is mixed degradation (e.g., blur plus noise or haze plus missing bands): because DP is continuous, it may interpolate between training families, but the paper only validates routing on single-type degradations.
  • Top-1 routing beating top-2 to top-4 suggests expert specialization is the active ingredient; a natural extension is to ask whether the DP metric space has an even smaller effective dimensionality than six, and whether two metrics could suffice.
  • Since the router is trained only on ARAD-labeled patches, a deployment on images from a very different sensor or platform could shift the metric distribution; recomputing the metric-selection pipeline on target-domain patches is a cheap calibration step before deployment.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper proposes DAMP, a unified hyperspectral image restoration framework that conditions a mixture-of-experts model on hand-crafted spatial-spectral metrics extracted directly from the degraded input. These six metrics (HFER, STU, SCM, SCSD, GSD, SCC) form a 'Degradation Prompt' (DP) that is used as the gating signal for a Degradation-Adaptive Mixture-of-Experts (DAMoE), whose experts are Spatial-Spectral Adaptive Modules (SSAMs). The model is trained jointly on Gaussian denoising, Gaussian deblurring, super-resolution, inpainting, and spectral band completion, using both natural (ARAD, ICVL) and remote sensing (PaviaU, PaviaC, Chikusei, Xiong'an, HyRank) datasets. The paper reports state-of-the-art average PSNR/SSIM in Table 2 on the five training-time tasks, and presents zero-shot results on motion deblurring and Poisson denoising on the unseen CAVE dataset (Table 3). Ablations in Section 4.3 show that DP contributes a large PSNR gain over a no-prompt baseline, that DP routing outperforms frequency-based and degradation-type routing, and that DP can be plugged into other architectures. The central claim is that DP enables a single label-free model to adaptively handle diverse, mixed, and unseen degradations without explicit priors.

Significance. If the central claim holds, DAMP would be a useful step toward unified HSI restoration without external degradation labels or text prompts. The paper has concrete strengths: (i) it proposes a testable, interpretable degradation descriptor instead of a learned black-box prompt; (ii) it provides a relatively broad evaluation across five restoration tasks and two domains; (iii) the overhead of the prompt is shown to be very small (Table 7, roughly +0.15 GFLOPs); and (iv) the authors release code, which supports reproducibility. The ablations (Tables 4-6, and Supplementary E) isolate the contributions of DP and SSAM and compare routing strategies, which is more than many papers in this area do. However, the paper's strongest generalization claim -- that the prompt transfers to genuinely unseen and mixed degradations -- is supported only by two same-family zero-shot experiments. The metric set was selected using labels from the five training-time degradation families, and no experiment evaluates a truly novel corruption mechanism or a mixture of degradations. Thus the significance depends on whether the authors can strengthen the generalization evidence; as written, the claim is plausib

major comments (4)
  1. [§4.2, Table 3] The zero-shot generalization claim rests on only two experiments: motion deblurring and Poisson denoising on CAVE. Both are within the same broad families as the training-time Gaussian blur and Gaussian noise. The abstract and Introduction claim robust restoration under 'mixed' degradations, but no experiment mixes two degradation types, and no experiment tests a genuinely new mechanism such as striping, compression artifacts, or haze. This is load-bearing: the DP router of Eq. (3) was trained and validated only on the five training families, so the transfer to 'unseen' degradations is not established. Please add experiments with mixed degradations (e.g., blur+noise, mask+noise) and/or out-of-family corruption types, and report the gating scores on unseen inputs to show that the router does not simply select an average expert.
  2. [Supplementary §A] The six metrics were selected from 25 candidates using a Random Forest trained on 500 ARAD patches labeled with the same five degradation types used in training. Thus the metric set is optimized to distinguish exactly the five training-time families. The claim that the same metric manifold will separate unseen physical mechanisms is not automatic; Fig. 2(c) even shows that different mechanisms can share similar metric distributions. Please provide an out-of-distribution analysis, e.g., project unseen degradation types into the DP space or report which experts are activated on them, to support the transfer claim. Without this, the 'general purpose descriptor' claim in §3.2 is stronger than the evidence.
  3. [Tables 2-3] All quantitative results are single-run point estimates with no error bars or statistical significance assessment. This is especially concerning for the remote sensing test sets, which are extremely small: HyRank uses two 128x128 random crops, and PaviaC/PaviaU each use one 256x256 random crop (Supplementary §B). Many differences in Table 2 and Table 3 are in the 0.1-0.5 dB range, which may be within run-to-run or crop-selection variation. Please report mean and standard deviation over at least three training seeds and/or multiple test crops, or otherwise justify why the reported differences are stable.
  4. [§3.3, Eq. (3)] The routing function is under-specified. The text says 'both the input x and the degradation prior e are fed into the routing function', but the method claims to avoid explicit degradation priors; it is not clear whether e is the DP vector, a projection of it, or something else. In addition, the top-k operator and Gaussian noise epsilon are described, but no load-balancing or auxiliary loss is mentioned, so it is possible that the router degenerates to always selecting the same expert. Please define e precisely, state whether any load-balancing loss is used, and include an analysis of routing behavior on training versus unseen degradations.
minor comments (5)
  1. [§3.2] The text states that three metrics (HFER, STU, SCM) are used, but then later states that six metrics form the DP. This should be reconciled; the first sentence appears to be an initial illustration, but it reads as a contradiction.
  2. [Table 1] STU is named 'Spatial Texture Uniformity' but its mathematical formulation is based on the geometric-to-arithmetic ratio of Fourier magnitudes, and the physical interpretation says 'Spectral smoothness'. This is confusing; please align the name, formula, and interpretation.
  3. [Figures 4-5] The label 'MP-HSIIR' appears in the figure panels; this should be 'MP-HSIR' to match the reference and text.
  4. [§4.1, Datasets] The data split description is ambiguous: 'we crop 1,000 images into patches' seems to refer to ARAD only, but the sentence also mentions ICVL. Please clarify how many patches are used per dataset and how the 100 non-overlapping test images are selected.
  5. [Eq. (3)] The notation uses e for what appears to be the DP vector, but elsewhere DP is denoted as a prompt. Please unify the notation and define e before Eq. (3).

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: DP is an input-only hand-crafted conditioning signal; the restoration network is trained end-to-end on external benchmarks, and the self-citations are not load-bearing.

full rationale

The DP is defined entirely by closed-form spatial-spectral statistics of the degraded input (Table 1/Table 8); it contains no ground-truth or restoration-output terms. The router in Eq. (3) uses the projected DP plus the input, and the restored image is produced by the SSAM/DAMoE experts via Eqs. (4)-(5), trained with an L1 loss against clean HSIs on ARAD/ICVL/remote-sensing data. The Random-Forest feature-importance analysis in Supp. A is an offline model-selection step on 500 ARAD patches: it selects which fixed formulas to use, but the selected six metrics remain deterministic functions of the input and are not re-fit to the target in the experiments. Table 5's ablation also argues against circularity: replacing DP routing with direct degradation-type routing lowers performance by 5.16 dB, so DP routing is not equivalent to using the training-time labels by construction. The self-citations to [16,18,19,59] are used for baseline/task context; the data-split protocol is attributed to external MP-HSIR [45]; none carries the central claim. The genuine weakness is evaluation coverage, not circularity: the zero-shot evidence (motion deblurring, Poisson denoising on CAVE) stays within the broad blur/noise families used in training, and the paper's own Fig. 2(c) shows that distinct degradation processes can share statistical signatures, so mixed-degradation behavior is not tested. That is a generalization/overclaim risk, not a definitional reduction of output to input, and it does not satisfy the hard-rule requirement of a specific equation-level or construction-level equivalence.

Assumptions & free parameters 5 free parameters · 4 assumptions · 2 invented entities

The system's gains rest on (i) a hand-distilled metric set chosen on the same benchmark family used for evaluation, (ii) hyperparameters (4 experts, top-1) tuned on that benchmark, and (iii) learned per-expert fusion weights whose claimed specialization is never measured. The only genuinely independent validation is the CAVE zero-shot pair, and even that pair stays within the noise/blur family of degradations.

free parameters (5)
  • Six-metric Degradation Prompt set (HFER, STU, SCM, SCSD, GSD, SCC) = Top-6 of 25 candidates by Random Forest importance on 500 labeled ARAD patches; pairwise Pearson redundancy cut rho<0.8
    The prompt representation is not derived from first principles; it is selected on a labeled validation subset of the same benchmark family (ARAD) used for the headline evaluation (Supplementary §A).
  • Metric-selection thresholds (rho<0.8; 7:3 and 80:20 splits) = rho = 0.8
    Hand-set redundancy threshold and train/test splits; no sensitivity analysis reported (Supplementary §A).
  • Total MoE experts and activated experts = 4 total, 1 activated (top-1)
    Chosen by ablation on the ARAD natural-data protocol (Tables 11-12 in supplement); 4/1 is the validation-best configuration, not a design necessity.
  • Spatial/spectral fusion weights lambda_s^(i), lambda_c^(i) = Learned per expert, constrained to sum to 1
    Eq. (6). Learned during training. The paper's qualitative claim about their learned specialization (blur -> low spatial weight) is asserted but never measured or reported.
  • Routing noise epsilon = Unspecified magnitude
    Gaussian noise introduced in Eq. (3) for load balancing; its magnitude is not reported, so the effective top-1 routing is not fully specified.
assumptions (4)
  • ad hoc to paper The six selected statistics form a complete-enough degradation descriptor that routing by them improves restoration beyond model capacity alone.
    §3.2 and Table 5 (+3.71 dB over frequency-based routing). This is the core inductive premise, validated only on the training task family and one zero-shot pair (CAVE).
  • domain assumption Random Forest feature importance on 500 labeled patches is a valid proxy for choosing prompt features.
    Supplementary §A. The selection is supervised by the same five degradation labels the model later handles, which weakens the 'label-free' framing.
  • domain assumption A U-Net/attention backbone trained with L1 loss and AdamW is a suitable base for unified HSI restoration.
    §3.3 and §4.1; inherited from prior protocols (PromptIR/MP-HSIR) without independent justification.
  • domain assumption PSNR/SSIM values are comparable across methods trained under potentially different protocols.
    §4.1-4.2. Comparisons against task-specific and unified baselines assume protocol-compatible splits and evaluation; single-run point estimates give no variance information.
invented entities (2)
  • Degradation Prompts (DP)
    purpose: Router input: a continuous multi-dimensional degradation descriptor intended to replace explicit degradation labels/text prompts.
    No falsifiable handle outside the paper: gains are shown only on ARAD (used in metric selection) and two CAVE zero-shot tasks of the same HSI modality. The two unseen tasks remain within the general 'noise/blur' family.
  • Spatial-Spectral Adaptive Modules (SSAM) / DAMoE experts
    purpose: Expert branches with learned spatial/spectral fusion weights, routed by DP.
    Architectural components; the claimed task specialization rests on in-paper ablations only, and the learned lambda values are never shown.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Degradation-Aware Metric Prompting for Hyperspectral Image Restoration." pith.science (2026). https://pith.science/paper/NX2TK4AW

@misc{pith2026251220251,
  author       = {Pith},
  title        = {Pith review of: Degradation-Aware Metric Prompting for Hyperspectral Image Restoration},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NX2TK4AW}},
  note         = {Machine review of arXiv:2512.20251}
}
read the original abstract

Unified hyperspectral image (HSI) restoration aims to recover diverse degradations within a single model. However, current methods often rely on impractical explicit priors or opaque black-box representations that overfit to training distributions, hampering generalization to unseen scenarios. To bridge this gap, we propose Degradation-Aware Metric Prompting (DAMP), a novel framework that characterizes multi-dimensional degradations through interpretable spatial-spectral metrics. These metrics serve as Degradation Prompts (DP), enabling the model to capture shared characteristics across tasks and adapt to unknown corruptions. Central to our framework is the Degradation-Adaptive Mixture-of-Experts (DAMoE), where Spatial-Spectral Adaptive Modules (SSAMs) serve as experts that utilize learnable fusion coefficients to specialize in distinct degradation degrees. By using DP as a gating router, DAMoE dynamically activates specialized experts tailored to the specific degradation profile. Extensive experiments on natural and remote sensing HSI datasets demonstrate that DAMP achieves state-of-the-art performance and exhibits exceptional zero-shot generalization on unseen restoration tasks. Code is publicly available at \href{DAMP}{https://github.com/MiliLab/DAMP}.

Figures

Figures reproduced from arXiv: 2512.20251 by the authors.

Figure 1
Figure 1. PSNR comparison with the state-of-the-art all-in-one [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. (a) Comparison between explicit prompt-based methods and degradation-aware metric prompting approaches. (b) Confusion [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. (a) The architecture of the proposed DAMP framework. (b) The Degradation-Adaptive MoE. [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Visual comparison of HSI recovery performance on tasks with known degradation types. From top to bottom: super-resolution on [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: Visual comparison of Poisson denoising on the CAVE [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Normalized digital number error across spectral bands [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 7
Figure 7. Figure 7: The ten most important metrics for distinguishing degra [PITH_FULL_IMAGE:figures/full_fig_p012_7.png]
Figure 8
Figure 8. Figure 8: Impact of total number of experts. Notably, significant gains are observed in image inpaint￾ing (+2.09dB) and spectral band completion (+0.90dB), while maintaining comparable SSIM values. These results indicate that DAMP effectively mitigates conflicts among different …
Figure 9
Figure 9. Figure 9: Impact of number of activated experts. at 1. As shown in [PITH_FULL_IMAGE:figures/full_fig_p015_9.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

61 extracted references · 4 linked inside Pith

  1. [1]

    Sparse recovery of hyper- spectral signal from natural rgb images

    Boaz Arad and Ohad Ben-Shahar. Sparse recovery of hyper- spectral signal from natural rgb images. InProceedings of the European Conference on Computer Vision, pages 19–34. Springer, 2016. 5, 6, 7, 12, 13

    2016

  2. [2]

    Ntire 2022 spectral recovery chal- lenge and data set

    Boaz Arad, Radu Timofte, Rony Yahel, Nimrod Morag, Amir Bernat, Yuanhao Cai, Jing Lin, Zudi Lin, Haoqian Wang, Yulun Zhang, et al. Ntire 2022 spectral recovery chal- lenge and data set. InProceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition, pages 863–881, 2022. 5, 6, 7, 12, 13, 14

    2022

  3. [3]

    Deep learning for classification of hyperspectral data: A comparative review.IEEE geoscience and remote sensing magazine, 7(2):159–173, 2019

    Nicolas Audebert, Bertrand Le Saux, and S ´ebastien Lef`evre. Deep learning for classification of hyperspectral data: A comparative review.IEEE geoscience and remote sensing magazine, 7(2):159–173, 2019. 1

    2019

  4. [4]

    Hsi-denet: Hyperspectral image restoration via convolutional neural network.IEEE Transactions on Geoscience and Remote Sensing, 57(2):667–682, 2018

    Yi Chang, Luxin Yan, Houzhang Fang, Sheng Zhong, and Wenshan Liao. Hsi-denet: Hyperspectral image restoration via convolutional neural network.IEEE Transactions on Geoscience and Remote Sensing, 57(2):667–682, 2018. 2

    2018

  5. [5]

    Mapping of forest alliances with simulated multi-seasonal hyperspectral satellite imagery.Remote Sensing of Environ- ment, 210:490–507, 2018

    Matthew L Clark, Jennifer Buck-Diaz, and Julie Evens. Mapping of forest alliances with simulated multi-seasonal hyperspectral satellite imagery.Remote Sensing of Environ- ment, 210:490–507, 2018. 1

    2018

  6. [6]

    In- structir: High-quality image restoration following human in- structions

    Marcos V Conde, Gregor Geigle, and Radu Timofte. In- structir: High-quality image restoration following human in- structions. InProceedings of the European Conference on Computer Vision, pages 1–21. Springer, 2024. 1, 2, 6, 7, 8

    2024

  7. [7]

    Total-variation-regularized low-rank matrix factoriza- tion for hyperspectral image restoration.IEEE transactions on geoscience and remote sensing, 54(1):178–188, 2015

    Wei He, Hongyan Zhang, Liangpei Zhang, and Huanfeng Shen. Total-variation-regularized low-rank matrix factoriza- tion for hyperspectral image restoration.IEEE transactions on geoscience and remote sensing, 54(1):178–188, 2015. 2

    2015

  8. [8]

    Hybrid convolutional and attention network for hyper- spectral image denoising.IEEE Geoscience and Remote Sensing Letters, 21:1–5, 2024

    Shuai Hu, Feng Gao, Xiaowei Zhou, Junyu Dong, and Qian Du. Hybrid convolutional and attention network for hyper- spectral image denoising.IEEE Geoscience and Remote Sensing Letters, 21:1–5, 2024. 6

    2024

Show all 61 references
  1. [9]

    Xin Huang and Liangpei Zhang. A comparative study of spa- tial approaches for urban mapping using hyperspectral rosis images over pavia city, northern italy.International Journal of Remote Sensing, 30(12):3205–3221, 2009. 5, 6, 7, 12, 13

    2009

  2. [10]

    Autodir: Automatic all-in-one image restoration with latent diffusion

    Yitong Jiang, Zhaoyang Zhang, Tianfan Xue, and Jinwei Gu. Autodir: Automatic all-in-one image restoration with latent diffusion. InProceedings of the European Conference on Computer Vision, pages 340–359. Springer, 2024. 2

    2024

  3. [11]

    Hyrank hyperspectral satellite dataset i, 2018

    Konstantinos Karantzalos, Christina Karakizi, Zacharias Kandylakis, and Georgia Antoniou. Hyrank hyperspectral satellite dataset i, 2018. 5, 6, 12, 13

    2018

  4. [12]

    Prompthsi: Universal hyperspectral image restoration with vision-language modulated frequency adaptation.arXiv preprint arXiv:2411.15922, 2024

    Chia-Ming Lee, Ching-Heng Cheng, Yu-Fan Lin, Yi-Ching Cheng, Wo-Ting Liao, Fu-En Yang, Yu-Chiang Frank Wang, and Chih-Chung Hsu. Prompthsi: Universal hyperspectral image restoration with vision-language modulated frequency adaptation.arXiv preprint arXiv:2411.15922, 2024. 1, 2, 3

    2024 arXiv

  5. [13]

    All-in-one image restoration for unknown corruption

    Boyun Li, Xiao Liu, Peng Hu, Zhongqin Wu, Jiancheng Lv, and Xi Peng. All-in-one image restoration for unknown corruption. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 17452– 17462, 2022. 2

    2022

  6. [14]

    Foundir: Unleashing million-scale training data to advance foundation models for image restoration.arXiv preprint arXiv:2412.01427, 2024

    Hao Li, Xiang Chen, Jiangxin Dong, Jinhui Tang, and Jin- shan Pan. Foundir: Unleashing million-scale training data to advance foundation models for image restoration.arXiv preprint arXiv:2412.01427, 2024. 2

    2024 arXiv

  7. [15]

    Ld-rps: Zero-shot uni- fied image restoration via latent diffusion recurrent posterior sampling.arXiv preprint arXiv:2507.00790, 2025

    Huaqiu Li, Yong Wang, Tongwen Huang, Hailang Huang, Haoqian Wang, and Xiangxiang Chu. Ld-rps: Zero-shot uni- fied image restoration via latent diffusion recurrent posterior sampling.arXiv preprint arXiv:2507.00790, 2025. 2

    2025

  8. [16]

    Spatial-spectral transformer for hyperspectral image denoising

    Miaoyu Li, Ying Fu, and Yulun Zhang. Spatial-spectral transformer for hyperspectral image denoising. InProceed- ings of the AAAI Conference on Artificial Intelligence, pages 1368–1376, 2023. 5, 6

    2023

  9. [17]

    Spatial-spectral transformer for hyperspectral image denoising

    Miaoyu Li, Ying Fu, and Yulun Zhang. Spatial-spectral transformer for hyperspectral image denoising. InProceed- ings of the AAAI Conference on Artificial Intelligence, pages 1368–1376, 2023. 1, 2

    2023

  10. [18]

    Spectral enhanced rectangle transformer for hyperspectral image denoising

    Miaoyu Li, Ji Liu, Ying Fu, Yulun Zhang, and Dejing Dou. Spectral enhanced rectangle transformer for hyperspectral image denoising. InProceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition, pages 5805–5814, 2023. 6

    2023

  11. [19]

    Latent diffusion enhanced rect- angle transformer for hyperspectral image restoration.IEEE Transactions on Pattern Analysis and Machine Intelligence,

    Miaoyu Li, Ying Fu, Tao Zhang, Ji Liu, Dejing Dou, Cheng- gang Yan, and Yulun Zhang. Latent diffusion enhanced rect- angle transformer for hyperspectral image restoration.IEEE Transactions on Pattern Analysis and Machine Intelligence,

  12. [20]

    Supervise-assisted self-supervised deep-learning method for hyperspectral image restoration.IEEE Transactions on Neu- ral Networks and Learning Systems, 36(4):7331–7344, 2024

    Miaoyu Li, Ying Fu, Tao Zhang, and Guanghui Wen. Supervise-assisted self-supervised deep-learning method for hyperspectral image restoration.IEEE Transactions on Neu- ral Networks and Learning Systems, 36(4):7331–7344, 2024. 6

    2024

  13. [21]

    Hyperspectral image super-resolution using deep convo- lutional neural network.Neurocomputing, 266:29–41, 2017

    Yunsong Li, Jing Hu, Xi Zhao, Weiying Xie, and JiaoJiao Li. Hyperspectral image super-resolution using deep convo- lutional neural network.Neurocomputing, 266:29–41, 2017. 1, 2

    2017

  14. [22]

    Sicong Liu, Daniele Marinelli, Lorenzo Bruzzone, and Francesca Bovolo. A review of change detection in multi- temporal hyperspectral images: Current techniques, applica- tions, and challenges.IEEE Geoscience and Remote Sensing Magazine, 7(2):140–158, 2019. 1

    2019

  15. [23]

    Hyperspectral image super- resolution via dual-domain network based on hybrid convo- lution.IEEE Transactions on Geoscience and Remote Sens- ing, 62:1–18, 2024

    Tingting Liu, Yuan Liu, Chuncheng Zhang, Liyin Yuan, Xiubao Sui, and Qian Chen. Hyperspectral image super- resolution via dual-domain network based on hybrid convo- lution.IEEE Transactions on Geoscience and Remote Sens- ing, 62:1–18, 2024. 6

    2024

  16. [24]

    Uhd-processer: Unified uhd image restora- tion with progressive frequency learning and degradation- aware prompts

    Yidi Liu, Dong Li, Xueyang Fu, Xin Lu, Jie Huang, and Zheng-Jun Zha. Uhd-processer: Unified uhd image restora- tion with progressive frequency learning and degradation- aware prompts. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 23...

    2025

  17. [25]

    Swin transformer: 9 Hierarchical vision transformer using shifted windows

    Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, and Baining Guo. Swin transformer: 9 Hierarchical vision transformer using shifted windows. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 10012–10022, 2021. 5

    2021

  18. [26]

    Dual self-attention swin transformer for hyperspectral image super-resolution.IEEE Transactions on Geoscience and Remote Sensing, 61:1–12, 2023

    Yaqian Long, Xun Wang, Meng Xu, Shuyu Zhang, Shuguo Jiang, and Sen Jia. Dual self-attention swin transformer for hyperspectral image super-resolution.IEEE Transactions on Geoscience and Remote Sensing, 61:1–12, 2023. 2

    2023

  19. [27]

    Visual-instructed degradation diffusion for all-in-one image restoration

    Wenyang Luo, Haina Qin, Zewen Chen, Libin Wang, Dan- dan Zheng, Yuming Li, Yufan Liu, Bing Li, and Weiming Hu. Visual-instructed degradation diffusion for all-in-one image restoration. InProceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition, pages...

    2025

  20. [28]

    Intriguing findings of frequency selection for image deblurring

    Xintian Mao, Yiming Liu, Fengze Liu, Qingli Li, Wei Shen, and Yan Wang. Intriguing findings of frequency selection for image deblurring. InProceedings of the AAAI Conference on Artificial Intelligence, pages 1905–1913, 2023. 6

    1905

  21. [29]

    Loformer: Local frequency transformer for im- age deblurring

    Xintian Mao, Jiansheng Wang, Xingran Xie, Qingli Li, and Yan Wang. Loformer: Local frequency transformer for im- age deblurring. InProceedings of the 32nd ACM Inter- national Conference on Multimedia, pages 10382–10391,

  22. [30]

    Rapid deter- mination of inner quality parameters of intact mango fruits using portable near infrared spectroscopy

    AA Munawar, R Hayati, and F Fachruddin. Rapid deter- mination of inner quality parameters of intact mango fruits using portable near infrared spectroscopy. InIOP Confer- ence Series: Earth and Environmental Science, page 012009,

  23. [31]

    Hir-diff: Unsupervised hyper- spectral image restoration via improved diffusion models

    Li Pang, Xiangyu Rui, Long Cui, Hongzhong Wang, Deyu Meng, and Xiangyong Cao. Hir-diff: Unsupervised hyper- spectral image restoration via improved diffusion models. In Proceedings of the IEEE/CVF Conference on Computer Vi- sion and Pattern Recognition, pages 3005–3014, 2024. 2

    2024

  24. [32]

    Promptir: Prompting for all-in- one image restoration.Advances in Neural Information Pro- cessing Systems, 36:71275–71293, 2023

    Vaishnav Potlapalli, Syed Waqas Zamir, Salman H Khan, and Fahad Shahbaz Khan. Promptir: Prompting for all-in- one image restoration.Advances in Neural Information Pro- cessing Systems, 36:71275–71293, 2023. 1, 2, 6, 7, 8

    2023

  25. [33]

    Yuntao Qian and Minchao Ye. Hyperspectral imagery restoration using nonlocal spectral-spatial structured sparse representation with noise estimation.IEEE Journal of Se- lected Topics in Applied Earth Observations and Remote Sensing, 6(2):499–515, 2012. 2

    2012

  26. [34]

    Gendeg: Diffusion-based degradation synthesis for generalizable all-in-one image restoration

    Sudarshan Rajagopalan, Nithin Gopalakrishnan Nair, Jay N Paranjape, and Vishal M Patel. Gendeg: Diffusion-based degradation synthesis for generalizable all-in-one image restoration. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 2814...

    2025

  27. [35]

    Wavelet-based sparse reduced-rank regression for hyperspectral image restoration.IEEE Transactions on Geo- science and Remote Sensing, 52(10):6688–6698, 2014

    Behnood Rasti, Johannes R Sveinsson, and Magnus Orn Ul- farsson. Wavelet-based sparse reduced-rank regression for hyperspectral image restoration.IEEE Transactions on Geo- science and Remote Sensing, 52(10):6688–6698, 2014. 2

    2014

  28. [36]

    Automatic hyperspectral image restoration using sparse and low-rank modeling.IEEE Geoscience and Re- mote Sensing Letters, 14(12):2335–2339, 2017

    Behnood Rasti, Magnus Orn Ulfarsson, and Pedram Ghamisi. Automatic hyperspectral image restoration using sparse and low-rank modeling.IEEE Geoscience and Re- mote Sensing Letters, 14(12):2335–2339, 2017. 2

    2017

  29. [37]

    Image restoration for remote sensing: Overview and toolbox.IEEE Geoscience and Remote Sens- ing Magazine, 10(2):201–230, 2021

    Behnood Rasti, Yi Chang, Emanuele Dalsasso, Loic Denis, and Pedram Ghamisi. Image restoration for remote sensing: Overview and toolbox.IEEE Geoscience and Remote Sens- ing Magazine, 10(2):201–230, 2021. 1

    2021

  30. [38]

    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. InInternational Conference on Medical image com- puting and computer-assisted intervention, pages 234–241. Springer, 2015. 4

    2015

  31. [39]

    Degradation-aware feature perturbation for all-in- one image restoration

    Xiangpeng Tian, Xiangyu Liao, Xiao Liu, Meng Li, and Chao Ren. Degradation-aware feature perturbation for all-in- one image restoration. InProceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition, pages 28165–28175, 2025. 2, 6, 7, 8

    2025

  32. [40]

    Stripformer: Strip transformer for fast image deblurring

    Fu-Jen Tsai, Yan-Tsung Peng, Yen-Yu Lin, Chung-Chi Tsai, and Chia-Wen Lin. Stripformer: Strip transformer for fast image deblurring. InProceedings of the European Confer- ence on Computer Vision, pages 146–162. Springer, 2022. 6

    2022

  33. [41]

    Hy- perspectral image super-resolution meets deep learning: A survey and perspective.IEEE/CAA Journal of Automatica Sinica, 10(8):1668–1691, 2023

    Xinya Wang, Qian Hu, Yingsong Cheng, and Jiayi Ma. Hy- perspectral image super-resolution meets deep learning: A survey and perspective.IEEE/CAA Journal of Automatica Sinica, 10(8):1668–1691, 2023. 2

    2023

  34. [42]

    Hyperspectral image restoration via total variation regularized low-rank tensor decomposition

    Yao Wang, Jiangjun Peng, Qian Zhao, Yee Leung, Xi-Le Zhao, and Deyu Meng. Hyperspectral image restoration via total variation regularized low-rank tensor decomposition. IEEE Journal of Selected Topics in Applied Earth Observa- tions and Remote Sensing, 11(4):1227–1243, 2017. 2

    2017

  35. [43]

    Image quality assessment: from error visibility to structural similarity.IEEE Transactions on Image Process- ing, 13(4):600–612, 2004

    Zhou Wang, Alan C Bovik, Hamid R Sheikh, and Eero P Si- moncelli. Image quality assessment: from error visibility to structural similarity.IEEE Transactions on Image Process- ing, 13(4):600–612, 2004. 6

    2004

  36. [44]

    Predictive filtering integrated generative remote sensing hyperspectral image inpainting.IEEE Geoscience and Remote Sensing Let- ters, 2025

    Yinhu Wu, Junping Zhang, and Dongyang Liu. Predictive filtering integrated generative remote sensing hyperspectral image inpainting.IEEE Geoscience and Remote Sensing Let- ters, 2025. 6

    2025

  37. [45]

    Mp- hsir: A multi-prompt framework for universal hyperspectral image restoration.arXiv preprint arXiv:2503.09131, 2025

    Zhehui Wu, Yong Chen, Naoto Yokoya, and Wei He. Mp- hsir: A multi-prompt framework for universal hyperspectral image restoration.arXiv preprint arXiv:2503.09131, 2025. 1, 2, 3, 5, 6, 7, 8, 12

    2025

  38. [46]

    Non-local similarity based tensor decomposition for hyperspectral image denoising

    Fan Xu, Xiao Bai, and Jun Zhou. Non-local similarity based tensor decomposition for hyperspectral image denoising. In 2017 IEEE International Conference on Image Processing (ICIP), pages 1890–1894. IEEE, 2017. 2

    2017

  39. [47]

    Generalized assorted pixel camera: post- capture control of resolution, dynamic range, and spectrum

    Fumihito Yasuma, Tomoo Mitsunaga, Daisuke Iso, and Shree K Nayar. Generalized assorted pixel camera: post- capture control of resolution, dynamic range, and spectrum. IEEE Transactions on Image Processing, 19(9):2241–2253,

  40. [48]

    Differential transformer.arXiv preprint arXiv:2410.05258, 2024

    Tianzhu Ye, Li Dong, Yuqing Xia, Yutao Sun, Yi Zhu, Gao Huang, and Furu Wei. Differential transformer.arXiv preprint arXiv:2410.05258, 2024. 5

    2024 arXiv

  41. [49]

    Aerial hyperspectral remote sensing classification dataset of xiongan new area (matiwan village).National Remote Sensing Bulletin, 24 (11):1299–1306, 2020

    CEN Yi, Lifu Zhang, Xia Zhang, W ANG Yueming, QI Wen- chao, TANG Senlin, and Peng Zhang. Aerial hyperspectral remote sensing classification dataset of xiongan new area (matiwan village).National Remote Sensing Bulletin, 24 (11):1299–1306, 2020. 5, 6, 7, 12, 13

    2020

  42. [50]

    Airborne hyperspectral data over chikusei.Space Appl

    Naoto Yokoya and Akira Iwasaki. Airborne hyperspectral data over chikusei.Space Appl. Lab., Univ. Tokyo, Tokyo, 10 Japan, Tech. Rep. SAL-2016-05-27, 5(5):5, 2016. 5, 6, 12, 13

    2016

  43. [51]

    V olformer: Explore more com- prehensive cube interaction for hyperspectral image restora- tion and beyond

    Dabing Yu and Zheng Gao. V olformer: Explore more com- prehensive cube interaction for hyperspectral image restora- tion and beyond. InProceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition, pages 28091–28101, 2025. 6

    2025

  44. [52]

    Dstrans: Dual-stream transformer for hyperspectral image restoration

    Dabing Yu, Qingwu Li, Xiaolin Wang, Zhiliang Zhang, Yixi Qian, and Chang Xu. Dstrans: Dual-stream transformer for hyperspectral image restoration. InProceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pages 3739–3749, 2023. 2

    2023

  45. [53]

    Qiangqiang Yuan, Qiang Zhang, Jie Li, Huanfeng Shen, and Liangpei Zhang. Hyperspectral image denoising employ- ing a spatial–spectral deep residual convolutional neural net- work.IEEE Transactions on Geoscience and Remote Sens- ing, 57(2):1205–1218, 2018. 2

    2018

  46. [54]

    Com- plexity experts are task-discriminative learners for any im- age restoration

    Eduard Zamfir, Zongwei Wu, Nancy Mehta, Yuedong Tan, Danda Pani Paudel, Yulun Zhang, and Radu Timofte. Com- plexity experts are task-discriminative learners for any im- age restoration. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages ...

    2025

  47. [55]

    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. 2

    2021

  48. [56]

    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,

  49. [57]

    Hyperspectral image restoration via cnn de- noiser prior regularized low-rank tensor recovery.Computer Vision and Image Understanding, 197:103004, 2020

    Haijin Zeng, Xiaozhen Xie, Haojie Cui, Yuan Zhao, and Jifeng Ning. Hyperspectral image restoration via cnn de- noiser prior regularized low-rank tensor recovery.Computer Vision and Image Understanding, 197:103004, 2020. 2

    2020

  50. [58]

    Hyperspectral image restoration using low-rank matrix recovery.IEEE transactions on geoscience and remote sensing, 52(8):4729–4743, 2013

    Hongyan Zhang, Wei He, Liangpei Zhang, Huanfeng Shen, and Qiangqiang Yuan. Hyperspectral image restoration using low-rank matrix recovery.IEEE transactions on geoscience and remote sensing, 52(8):4729–4743, 2013. 1

    2013

  51. [59]

    Essaformer: Efficient transformer for hyperspectral image super-resolution

    Mingjin Zhang, Chi Zhang, Qiming Zhang, Jie Guo, Xinbo Gao, and Jing Zhang. Essaformer: Efficient transformer for hyperspectral image super-resolution. InProceedings of the IEEE/CVF International Conference on Computer Vision, pages 23073–23084, 2023. 5, 6

    2023

  52. [60]

    Selective hourglass mapping for universal image restoration based on diffusion model

    Dian Zheng, Xiao-Ming Wu, Shuzhou Yang, Jian Zhang, Jian-Fang Hu, and Wei-Shi Zheng. Selective hourglass mapping for universal image restoration based on diffusion model. InProceedings of the IEEE/CVF conference on com- puter vision and pattern recognition, pages 25445–25455,

  53. [61]

    Unires: Universal image restoration for complex degradations.arXiv preprint arXiv:2506.05599, 2025

    Mo Zhou, Keren Ye, Mauricio Delbracio, Peyman Milan- far, Vishal M Patel, and Hossein Talebi. Unires: Universal image restoration for complex degradations.arXiv preprint arXiv:2506.05599, 2025. 2 11 This supplementary material provides more details and results that are not inc...

    2025 arXiv

Pith tools

Reviewed August 3, 2026 · model on record in the stance chip above.