Pith. sign in

REVIEW 4 major objections 14 minor 1 cited by

ResPanDiff: Diffusion Model for Pansharpening by Inferring Residual Inference

T0 review · 4 major / 14 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read ResPanDiff claims a 15-step diffusion process for pansharpening, cutting sampling steps by over 90 percent while keeping or improving fusion quality on three satellite datasets.

desk verdict A solid engineering contribution that applies residual-shift diffusion to pansharpening, but the state-of-the-art margin is within noise and the 15-step efficiency story is backed by qualitative plots rather than a measured geometric quantity. read the letter →

arxiv 2501.05091 v2 pith:XHVEL5JE submitted 2025-01-09 cs.CV eess.IV

Shiqi Cao , Liangjian Deng , Shangqi Deng This is my paper · ORCID
classification cs.CVeess.IV
keywords pansharpeningdiffusionmodelresidualinferenceimagefusionacceleratedsamplingMarkovchainremotesensingmulti-spectral
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 make diffusion-based pansharpening fast enough for practical use by generating the residual between the high-resolution multispectral target and the low-resolution multispectral input, rather than generating the full image. The authors build a Markov chain whose endpoints are a noisy residual and the true residual $e_0 = \mathrm{HRMS} - \mathrm{LRMS}$, and they sample only 15 diffusion steps instead of the hundreds used by standard baselines. They report state-of-the-art quality on the WorldView-3, GaoFen-2, and QuickBird datasets, with ablations showing that the latent input state, the shallow conditional injection, and the custom loss each contribute to the gain. If the claim holds, slow inference stops being the main obstacle to diffusion models in satellite image fusion.

What carries the argument

The load-bearing object is a residual-transport Markov chain: the forward transition $q(e_t|e_{t-1},e_0)$ drifts the noisy residual toward zero with step size $\alpha_t$, giving the closed-form marginal $q(e_t|e_0)=\mathcal{N}((1-\bar{\alpha}_t)e_0,\kappa^2\bar{\alpha}_t)$ with $\bar{\alpha}_t=\sum_{i=1}^t\alpha_i$. The reverse transition $q(e_{t-1}|e_t,e_0)$ is a tractable Gaussian whose mean combines the noisy residual and the predicted residual, so training reduces to minimizing $\|f_\theta(x_t,c,t)-e_0\|_2^2$. The model receives $x_t=e_t+\mathrm{LRMS}$ as input, injects LRMS/PAN conditions through shallow conditional modulation, and is trained with a custom residual loss plus a boundary penalty that keeps predicted residuals inside the observed residual range. This chain is what allows 15-step sampling: because the transport is nearly straight, a short recursion over the residual is enough.

What would settle it

Run the trained ResPanDiff on a WV3 test set with both 15 and 50 sampling steps, and also compute the average angle between successive velocity estimates along the sampled trajectories. If the 50-step version improves SAM or ERGAS well beyond the reported standard-deviation margins, or if the trajectories show substantial curvature or crossing, the straight-path assumption behind the 15-step claim is not holding.

Watch

Extended reading notes

Core claim

The central claim is that the transport from LRMS to HRMS in pansharpening is almost straight, so a short diffusion chain can traverse it accurately if the diffusion process is defined on residuals instead of full images. Concretely, the paper sets the target as $e_0 = \mathrm{HRMS} - \mathrm{LRMS}$ and defines a forward process $q(e_t|e_0) = \mathcal{N}((1-\bar{\alpha}_t)e_0, \kappa^2\bar{\alpha}_t)$, where $\bar{\alpha}_t$ accumulates the schedule, so the chain ends at essentially noise. The reverse process is derived from Bayes' theorem as a Gaussian posterior, and the model is trained to predict $e_0$ from the latent state $x_t = e_t + \mathrm{LRMS}$ with LRMS and PAN supplied through wavelet-decomposed conditional modulation. At inference the chain runs for 15 steps, and the final HRMS is obtained by adding the predicted residual to LRMS. The reported experiments show this residual-inference chain beating previous diffusion and non-diffusion methods on WV3, GF2, and QB, including a zero-shot transfer from WV3 to WV2.

Load-bearing premise

The 15-step speedup rests on the empirical claim, supported only by qualitative trajectory plots, that the path from low-resolution multispectral input to high-resolution multispectral target is nearly straight with almost no crossing paths; if that path is actually curved or tangled, the short chain will not track it accurately.

Editorial extensions

If this is right

  • Diffusion pansharpening becomes practical at roughly 15 steps instead of hundreds, with reported quality above the previous diffusion state of the art on all three benchmark datasets.
  • The residual-inference design gives a template for other image-restoration tasks: define the diffusion target as the input–output difference and start the chain from a noisy residual rather than pure Gaussian noise.
  • The custom loss with boundary penalty avoids the small-gradient and overfitting behavior the paper observes with $\ell^2$ loss, so training continues improving instead of stalling.
  • No two-stage reflow training is needed for pansharpening; a single training pass produces a nearly straight transport path, so the usual Rectified Flow reflow cost is bypassed.
  • Zero-shot results on WorldView-2 suggest a model trained on one satellite sensor can be applied to another without retraining.

Reading between the lines

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

  • A quantitative straightness metric—mean cosine between successive velocity vectors, or a crossing count along sampled trajectories—would let future work predict the minimum safe step count for a new fusion dataset before training; the paper itself only shows qualitative plots.
  • The boundary penalty with weight 10000 effectively hard-clips predicted residuals to the training distribution's range, so on sensors with very different radiometry the penalty may suppress valid detail; testing on additional sensors with QNR and Dλ would reveal the limit.
  • The same Markov chain should apply to hyperspectral pansharpening or to sharpening-plus-denoising, where the residual is even sparser and the straight-path assumption may hold more strongly; a direct test would be to run ResPanDiff on hyperspectral fusion benchmarks.
  • Since the schedule is fixed at T=15, a learned or per-dataset schedule might reduce the step count further or recover quality on harder scenes; that is a natural next experiment not reported in the paper.
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 / 14 minor

Summary. The paper proposes ResPanDiff, a diffusion model for pansharpening that targets the residual between HRMS and LRMS images rather than the full HRMS image. The forward process adds noise to the residual, and the reverse process denoises it conditioned on the LRMS and PAN inputs. The authors argue that the transport path from LRMS to HRMS is nearly straight, which allows them to reduce sampling to 15 steps. The method also introduces a latent-state input, a shallow condition-injection module, and a custom piecewise loss with a boundary penalty. Experiments on WV3, GF2, and QB datasets report metrics that are competitive with or better than recent baselines, and ablations are used to support each proposed component.

Significance. If the derivation and empirical claims are both solid, the paper would make a useful contribution to accelerated diffusion-based pansharpening: modeling the residual as the direct diffusion target is an intuitive idea that connects with residual-shifting methods, and a 15-step sampler would be a practical speedup. The architectural components (latent state, shallow condition injection, custom loss) are plausible and could transfer to other fusion tasks. However, the paper's central speedup claim rests on an unquantified geometric premise, the reverse-process formulas contain mathematical inconsistencies, and the reported improvements over the main baseline DDIF are within one standard deviation with no significance testing. The manuscript does not include code, which further limits verification. With corrected derivations, clearer evaluation of the step-reduction claim, and stronger statistical support, the work could meet the bar for publication.

major comments (4)
  1. [Section III-A, Eq. (17), Appendix B] The posterior mean in Eq. (17) is incorrect for the forward process defined in Eq. (10). With the forward increment δ_t and cumulative sum α_t = Σ_{i≤t} δ_i, the posterior q(et−1|et,e0) has mean (α_{t−1}/α_t) et + (δ_t/α_t) e0, not (α_{t−1}/α_t) et + e0. The derivation in Eqs. (39)–(43) drops the increment factor: μ1 should be et + δ_t e0, and the final combination should yield a coefficient δ_t/α_t on e0, not α_t/α_t = 1. As written, the reverse update in Algorithm 2 adds the full residual e0 at every step, which is not the correct posterior and would be expected to destabilize sampling. This is a load-bearing mathematical error in the method definition, not a cosmetic typo; the authors must correct the formulas and re-verify the sampling behavior, or explicitly justify a schedule under which their simplified mean is valid.
  2. [Algorithm 1 and Algorithm 2] Both pseudocode listings contain errors that block reproducibility. Algorithm 1, step 6, writes et ← (1 − αt)e0 + καt, which is deterministic and dimensionally inconsistent; it should be et = (1 − α_t)e0 + κ√α_t ξ with ξ ∼ N(0,I). Algorithm 2, step 6, updates xt ← et + xT using the old residual et instead of the newly sampled et−1; it should be xt ← et−1 + xT. Additionally, the initialization of eT is not specified in Algorithm 2; sampling eT from q(eT|e0) would require e0, which is unknown at inference, so the paper must state the actual prior used to start the reverse chain.
  3. [Section V-C, Figs. 6 and 14, Section IV-C] The central 15-step acceleration claim is not rigorously supported. The straightness of the LRMS-to-HRMS transport is evidenced only by 2D scatter projections in Figs. 6 and 14, with no description of the embedding and no quantitative crossing statistic. Furthermore, the reverse procedure in Eq. (17) is a stochastic ancestral sampler, not the deterministic ODE Euler solver for which straightness would directly justify large step sizes. The paper should provide a quantitative measure of path crossings (e.g., the fraction of trajectory pairs that intersect in the full state space) and an analysis of discretization error for the actual stochastic sampler at 15 steps. In addition, Table V compares DDIM and ResShift at 15 steps but not DDIF at 15 steps, so the step reduction relative to the main baseline is not isolated.
  4. [Tables II, III, IV] The claimed state-of-the-art performance is not statistically substantiated. On WV3, the SAM improvement over DDIF is 2.7179±0.4921 vs. 2.7386±0.5080, which is well within one standard deviation; on QB, the ERGAS values are 3.4937±1.3837 vs. 3.5223±0.2703, again overlapping. No significance tests or confidence intervals are reported for any of the benchmark comparisons. The paper should add paired significance tests across the 20 test samples (e.g., paired t-test or Wilcoxon signed-rank test) or temper the SOTA claim to 'comparable performance.'
minor comments (14)
  1. [Title page] The author list contains 'Shangqi Deng 1' twice; this duplicate should be removed.
  2. [Abstract] The abstract contains a sentence fragment: after 'loss functions to give a better guidance for the residual generation task,' the next sentence begins with a lowercase 'enabling' and should be merged or rewritten.
  3. [Section III-A] 'staring at eT' should be 'starting at eT'.
  4. [Eq. (1)] The phrase 'the following OEDs' should be 'the following ODEs'.
  5. [Notation, Eq. (10) vs Eq. (11)] The symbol α_t is used both for the per-step increment and for the cumulative sum Σ_{i≤t} α_i, which causes confusion in Eq. (17) and Eq. (18); the authors should introduce separate notation (e.g., δ_t for the increment and α_t for the cumulative sum).
  6. [Algorithm 1] In addition to the missing noise term, step 6 also omits the square root on α_t; the reparameterization should be et = (1 − α_t)e0 + κ√α_t ξ.
  7. [Eq. (39)] In Appendix B, μ1 = et + α_t should be μ1 = et + δ_t e0, where δ_t is the per-step increment.
  8. [Fig. 10] The caption refers to a parameter 'p' that is likely 'α' and should be clarified; the x-axis labels are also not readable.
  9. [Figs. 12 and 13] The axis labels contain garbled placeholder text (e.g., '/uni0000004c/uni00000057/...') and should be replaced with proper mathematical notation.
  10. [Section III-D, Eq. (23)] The boundary penalty Lp is not fully specified: it should be stated whether max(e0) and min(e0) are computed per image, per patch, or globally, and how the clamp operation Fc handles negative arguments.
  11. [Eq. (22)] The constants a and b in the residual loss are given without derivation; the authors should show the continuity and differentiability conditions that determine them.
  12. [References] References [29] and [32] are both DPM-Solver and should be merged into a single entry.
  13. [Table V] The loss-function ablation is performed on GF2 while the other ablations are on WV3, which makes the magnitude of the improvements not directly comparable; this should be stated explicitly or the ablations should be run on a common dataset.
  14. [Section IV-C] The paper claims 'over 90% step compared with the benchmark diffusion models,' but it does not state the number of sampling steps used by PanDiff or DDIF; the actual step counts should be reported to support this percentage.

Circularity Check

1 steps flagged · score 3.0 of 10

Residual diffusion is a genuinely supervised, externally benchmarked method; the only by-construction element is the straight-trajectory justification for the 15-step speedup, plus a minor, non-load-bearing self-citation.

  1. self definitional [Sec. III-A (Why does residual generation make sense?), Eq. (11) with e0 = X1 - X0; Figs. 6 and 14; Sec. V-C]
    "Based on this assumption, we plot the distribution of WV3 data and the generation path from LRMS to HRMS, and our hypothesis has been confirmed... In Fig. 6, we can find that the transport trajectory from LRMS to HRMS is straight with minimal cross-interference. This suggests that in the MSIF task, we do not need to go through stages like the 1-Rectified flow to avoid cross-interference; instead, we can find a straightforward generation path."

    With e0 defined as X1 - X0, the forward-process marginal in Eq. (11) gives x_t = x_T + (1 - alpha-bar_t)e0 = (1 - alpha-bar_t)X1 + alpha-bar_t X_T, which is exactly the Rectified-Flow linear interpolation. The plotted trajectory is therefore straight by construction for every sample, not an empirical discovery that residual inference makes the path straight. The residual-justifies-15-steps argument consequently rests on the unquantified part, 'minimal cross-interference', shown only as 2D scatter projections in Figs. 6 and 14. The 15-step performance itself, however, is an externally benchmarked experimental result and does not reduce to this geometric premise.

full rationale

The central training objective is a standard supervised regression to the residual e0 = HRMS - LRMS (Eq. 20), and the reverse posterior in Eq. (17) follows from Bayes' theorem and the stated forward process; the Appendix derivation is algebraically consistent. Performance is measured on external WV3, GF2, and QB test sets, so there is no fitted-input-called-prediction loop and no statistical forcing of the headline accuracy numbers. The main circularity-adjacent element is motivational: the paper claims that residual inference yields a straight transport path and hence allows 15 steps, but the straightness is built into the definition of e0 and the linear forward marginal (Eq. 11); only the claimed 'minimal cross-interference' is genuinely empirical, and it is not quantified. Also, the architecture borrows CSM/FWM from DDIF [26], a paper with overlapping authorship (ShiQi Cao and Liangjian Deng), but this component is ablated and is not the source of the speedup claim; it is a minor self-citation rather than a load-bearing circular step. Overall, the residual-diffusion derivation is self-contained and externally validated, while the step-count justification contains one by-construction geometric claim, warranting a low-moderate circularity score.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The central efficiency claim depends on a small number of hand-selected hyperparameters (T=15, kappa=1, gamma=10000, cosine schedule parameter) and on two domain assumptions: the residual transport path is straight, and the terminal residual prior carries no conditioning information. The loss-shape constants a and b are derived from continuity, not fit. No new entities are postulated.

free parameters (5)
  • T (number of diffusion steps) = 15
    Chosen to balance speed and quality; the paper claims 15 steps reduce over 90% of steps, but no systematic search over T is shown.
  • kappa (noise variance scale) = 1
    Controls forward-process noise variance in Eq (10); fixed to 1 in the Fig 10 experiments.
  • gamma (boundary penalty weight) = 10000
    Set empirically in Eq (24); the paper says this value gives evident improvement but does not report a sweep.
  • noise schedule parameter in cosine schedule = not stated in text (range 8e-3 to 8e-1 tested)
    Eq (25) has a hyperparameter controlling the noise schedule; Fig 10 shows tests but the final value is not reported.
  • loss shape constants a and b = a=1/(2e)-1/2, b=7/4-3/(2e)-1/(4e^2)
    Derived from requiring the piecewise loss in Eq (22) to be continuous and C1 at |h|=1; not fitted to data, but the functional form is hand-chosen.
assumptions (4)
  • domain assumption Rectified flow theory: straight transport paths allow accurate ODE simulation with few Euler steps.
    Invoked in Sec III-A to argue that residual generation gives straight paths and thus fewer steps; external theory from [33].
  • domain assumption Transport trajectory from LRMS to HRMS is nearly straight with minimal crossovers for pansharpening data.
    Assumed in Sec III-A and validated only by qualitative plots (Fig 6, Fig 14); the entire 15-step efficiency claim rests on this.
  • standard math The prior p(eT|xT) approaches zero (Eq 12).
    Assumes the terminal noisy residual is pure noise independent of the conditioning image, consistent with the variance schedule.
  • domain assumption LRMS and HRMS can be assumed to share spatial resolution after pre-upsampling.
    Stated in Sec III before the notations; needed so that the residual is pixel-wise defined.

how reviews work

0 comments
Cite this review

Pith. "Pith review of ResPanDiff: Diffusion Model for Pansharpening by Inferring Residual Inference." pith.science (2026). https://pith.science/paper/XHVEL5JE

@misc{pith2026250105091,
  author       = {Pith},
  title        = {Pith review of: ResPanDiff: Diffusion Model for Pansharpening by Inferring Residual Inference},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XHVEL5JE}},
  note         = {Machine review of arXiv:2501.05091}
}
abstract

The implementation of diffusion-based pansharpening task is predominantly constrained by its slow inference speed, which results from numerous sampling steps. Despite the existing techniques aiming to accelerate sampling, they often compromise performance when fusing multi-source images. To ease this limitation, we introduce a novel and efficient diffusion model named Diffusion Model for Pansharpening by Inferring Residual Inference (ResPanDiff), which significantly reduces the number of diffusion steps without sacrificing the performance to tackle pansharpening task. In ResPanDiff, we innovatively propose a Markov chain that transits from noisy residuals to the residuals between the LRMS and HRMS images, thereby reducing the number of sampling steps and enhancing performance. Additionally, we design the latent space to help model extract more features at the encoding stage, Shallow Cond-Injection~(SC-I) to help model fetch cond-injected hidden features with higher dimensions, and loss functions to give a better guidance for the residual generation task. enabling the model to achieve superior performance in residual generation. Furthermore, experimental evaluations on pansharpening datasets demonstrate that the proposed method achieves superior outcomes compared to recent state-of-the-art~(SOTA) techniques, requiring only 15 sampling steps, which reduces over $90\%$ step compared with the benchmark diffusion models. Our experiments also include thorough discussions and ablation studies to underscore the effectiveness of our approach.

Figures

Figures reproduced from arXiv: 2501.05091 by the authors.

Figure 1
Figure 1. (a) Pansharpening involves fusing the PAN and LRMS [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. The score of Frechet Inception Distance (FID [30]), [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. The different lines the transport distribution [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (11 more)
Figure 4
Figure 4. Figure 4: The utility of Rectified flow. We can find that the [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 5
Figure 5. Figure 5: The different stages of the Rectified flow. The first [PITH_FULL_IMAGE:figures/full_fig_p004_5.png]
Figure 7
Figure 7. Figure 7: The overview of the proposed diffusion process. It builds up a Markov chain between the [PITH_FULL_IMAGE:figures/full_fig_p006_7.png]
Figure 8
Figure 8. Figure 8: Overview of the architecture of the model. The input gets [PITH_FULL_IMAGE:figures/full_fig_p006_8.png]
Figure 10
Figure 10. Figure 10: It can be observed that as the value of α increases, the images tend to contain higher levels of noise [PITH_FULL_IMAGE:figures/full_fig_p007_10.png]
Figure 9
Figure 9. Figure 9: The left one represents the figs of different losses. Comparing with [PITH_FULL_IMAGE:figures/full_fig_p008_9.png]
Figure 10
Figure 10. Figure 10: Illustration of the noise schedule for ResPanDiff. We [PITH_FULL_IMAGE:figures/full_fig_p008_10.png]
Figure 11
Figure 11. Figure 11: Visual comparisons with previous pansharpening methods on the WorldView-3 dataset. The first and third rows show [PITH_FULL_IMAGE:figures/full_fig_p009_11.png]
Figure 12
Figure 12. Figure 12: SAM on different loss. We can find that at the [PITH_FULL_IMAGE:figures/full_fig_p010_12.png]
Figure 13
Figure 13. Figure 13: Grad on different losses. We can observe that the [PITH_FULL_IMAGE:figures/full_fig_p010_13.png]
Figure 14
Figure 14. Figure 14: The first column represents the data distribution of [PITH_FULL_IMAGE:figures/full_fig_p010_14.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Two-Stage Random Alternation Framework for One-Shot Pansharpening

    cs.CV 2025-05 conditional novelty 6.0 of 10

    A one-shot pansharpening method that alternates random low- and full-resolution training per image pair, with a warm-up phase, achieves state-of-the-art full-resolution fusion quality on WV3, QB, and GF2 datasets.

Reference graph

Works this paper leans on

51 extracted references · 34 canonical work pages · cited by 1 Pith paper

  1. [1]

    Extracting spectral contrast in landsat thematic mapper image data using selective principal component analysis,

    P. S. Chavez and A. Y . Kwarteng, “Extracting spectral contrast in landsat thematic mapper image data using selective principal component analysis,” Photogrammetric Engineering and Remote Sensing , vol. 55, pp. 339–348, 1989

    work page 1989

  2. [2]

    Process for enhancing the spatial resolution of multispectral imagery using pan-sharpening,

    C. A. Laben and B. V . Brower, “Process for enhancing the spatial resolution of multispectral imagery using pan-sharpening,” Jan. 4 2000, uS Patent 6,011,875

    2000

  3. [3]

    Improving component substitution pansharpening through multivariate regression of ms + pan data,

    B. Aiazzi, S. Baronti, and M. Selva, “Improving component substitution pansharpening through multivariate regression of ms + pan data,” IEEE Transactions on Geoscience and Remote Sensing , vol. 45, no. 10, pp. 3230–3239, 2007

    2007

  4. [4]

    Hyperspectral pansharpening with guided filter,

    J. Qu, Y . Li, and W. Dong, “Hyperspectral pansharpening with guided filter,” IEEE Geoscience and Remote Sensing Letters , vol. 14, no. 11, pp. 2152–2156, 2017

    2017

  5. [5]

    Introduction of sensor spectral response into image fusion methods. application to wavelet-based methods,

    X. Otazu, M. Gonz ´alez-Aud´ıcana, O. Fors, and J. N ´u˜nez, “Introduction of sensor spectral response into image fusion methods. application to wavelet-based methods,” IEEE Transactions on Geoscience and Remote Sensing, vol. 43, no. 10, pp. 2376–2385, 2005

    2005

  6. [6]

    Smoothing filter-based intensity modulation: A spectral preserve image fusion technique for improving spatial details,

    J. Liu, “Smoothing filter-based intensity modulation: A spectral preserve image fusion technique for improving spatial details,” International Journal of Remote Sensing , vol. 21, no. 18, pp. 3461–3472, 2000

    work page 2000

  7. [7]

    MTF- tailored multiscale fusion of high-resolution ms and pan imagery,

    B. Aiazzi, L. Alparone, S. Baronti, A. Garzelli, and M. Selva, “MTF- tailored multiscale fusion of high-resolution ms and pan imagery,” Photogrammetric Engineering and Remote Sensing , vol. 72, no. 5, pp. 591–596, 2006

    work page 2006

  8. [8]

    Contrast and error-based fusion schemes for multispectral image pansharpening,

    G. Vivone, R. Restaino, M. Dalla Mura, G. Licciardi, and J. Chanus- sot, “Contrast and error-based fusion schemes for multispectral image pansharpening,” IEEE Geoscience and Remote Sensing Letters , vol. 11, no. 5, pp. 930–934, 2013

    work page 2013

Show all 51 references
  1. [9]

    A new pansharpening method based on spatial and spectral sparsity priors,

    X. He, L. Condat, J. M. Bioucas-Dias, J. Chanussot, and J. Xia, “A new pansharpening method based on spatial and spectral sparsity priors,” IEEE Transactions on Image Processing, vol. 23, no. 9, pp. 4160–4174, 2014

    2014

  2. [10]

    A variational approach to hyperspectral image fusion,

    M. Moeller, T. Wittman, and A. L. Bertozzi, “A variational approach to hyperspectral image fusion,” in Algorithms and Technologies for Multispectral, Hyperspectral, and Ultraspectral Imagery XV , vol. 7334. SPIE, 2009, pp. 502–511

    2009

  3. [11]

    High-quality bayesian pan- sharpening,

    T. Wang, F. Fang, F. Li, and G. Zhang, “High-quality bayesian pan- sharpening,” IEEE Transactions on Image Processing , vol. 28, no. 1, pp. 227–239, 2018

    2018

  4. [12]

    Lrtcfpan: Low-rank tensor completion based framework for pansharpening,

    Z.-C. Wu, T.-Z. Huang, L.-J. Deng, J. Huang, J. Chanussot, and G. Vivone, “Lrtcfpan: Low-rank tensor completion based framework for pansharpening,” IEEE Transactions on Image Processing , vol. 32, pp. 1640–1655, 2023

    2023

  5. [13]

    Pansharpening by convolutional neural networks,

    G. Masi, D. Cozzolino, L. Verdoliva, and G. Scarpa, “Pansharpening by convolutional neural networks,” Remote Sensing , vol. 8, no. 7, p. 594, 2016

    2016

  6. [14]

    PanNet: A deep network architecture for pan-sharpening,

    J. Yang, X. Fu, Y . Hu, Y . Huang, X. Ding, and J. W. Paisley, “PanNet: A deep network architecture for pan-sharpening,” in Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) , 2017, pp. 1753–1761

    2017

  7. [15]

    Detail injection- based deep convolutional neural networks for pansharpening,

    L.-J. Deng, G. Vivone, C. Jin, and J. Chanussot, “Detail injection- based deep convolutional neural networks for pansharpening,” IEEE Transactions on Geoscience and Remote Sensing , vol. 59, no. 8, pp. 6995–7010, 2020

    2020

  8. [16]

    Model- informed multistage unsupervised network for hyperspectral image super-resolution,

    J. Li, K. Zheng, L. Gao, L. Ni, M. Huang, and J. Chanussot, “Model- informed multistage unsupervised network for hyperspectral image super-resolution,” IEEE Transactions on Geoscience and Remote Sens- ing, vol. 62, pp. 1–17, 2024

    2024

  9. [17]

    Deep residual learning for image recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , 2016, pp. 770–778

    2016

  10. [18]

    U-net: Convolutional networks for biomedical image segmentation,

    O. Ronneberger, P. Fischer, and T. Brox, “U-net: Convolutional networks for biomedical image segmentation,” in Medical Image Computing and Computer-Assisted Intervention (MICCAI) . Springer, 2015, pp. 234– 241

    2015

  11. [19]

    Remote sensing image fusion based on two-stream fusion network,

    X. Liu, Q. Liu, and Y . Wang, “Remote sensing image fusion based on two-stream fusion network,” Information Fusion, vol. 55, pp. 1–15, 2020

    2020

  12. [20]

    Denoising diffusion probabilistic models,

    J. Ho, A. Jain, and P. Abbeel, “Denoising diffusion probabilistic models,” in Advances in Neural Information Processing Systems (NeurIPS) , vol. 33, 2020, pp. 6840–6851

    2020

  13. [21]

    Elucidating the design space of diffusion-based generative models,

    T. Karras, M. Aittala, T. Aila, and S. Laine, “Elucidating the design space of diffusion-based generative models,” in Advances in Neural Information Processing Systems (NeurIPS) , vol. 35, 2022, pp. 26 565– 26 577

    2022

  14. [22]

    Palette: Image-to-image diffusion models,

    C. Saharia, W. Chan, H. Chang, C. Lee, J. Ho, T. Salimans, D. Fleet, and M. Norouzi, “Palette: Image-to-image diffusion models,” in ACM SIGGRAPH 2022 Conference Proceedings (ACM SIGGRAPH) , 2022, pp. 1–10

    2022

  15. [23]

    Image Super-Resolution via Iterative Refinement,

    C. Saharia, J. Ho, W. Chan, T. Salimans, D. J. Fleet, and M. Norouzi, “Image Super-Resolution via Iterative Refinement,” arXiv e-prints , p. arXiv:2104.07636, Apr. 2021

    2021 arXiv

  16. [24]

    Resshift: Efficient diffusion model for image super-resolution by residual shifting,

    Z. Yue, J. Wang, and C. C. Loy, “Resshift: Efficient diffusion model for image super-resolution by residual shifting,” Advances in Neural Information Processing Systems , vol. 36, 2024

    2024

  17. [25]

    Pandiff: A novel pansharpening method based on denoising diffusion probabilistic model,

    Q. Meng, W. Shi, S. Li, and L. Zhang, “Pandiff: A novel pansharpening method based on denoising diffusion probabilistic model,” IEEE Trans- actions on Geoscience and Remote Sensing , vol. 61, pp. 1–17, 2023

    2023

  18. [26]

    Diffusion model with disentangled modulations for sharpening multispectral and hyperspectral images,

    Z. Cao, S. Cao, L.-J. Deng, X. Wu, J. Hou, and G. Vivone, “Diffusion model with disentangled modulations for sharpening multispectral and hyperspectral images,” Information Fusion, vol. 104, p. 102158, 2024

    2024

  19. [27]

    Denoising diffusion implicit models,

    J. Song, C. Meng, and S. Ermon, “Denoising diffusion implicit models,” in International Conference on Learning Representations (ICLR) , 2021

    2021

  20. [28]

    Improved denoising diffusion probabilis- tic models,

    A. Q. Nichol and P. Dhariwal, “Improved denoising diffusion probabilis- tic models,” in International Conference on Machine Learning (ICML) . PMLR, 2021, pp. 8162–8171

    2021

  21. [29]

    Dpm-solver: A fast ode solver for diffusion probabilistic model sampling in around 10 steps,

    C. Lu, Y . Zhou, F. Bao, J. Chen, C. Li, and J. Zhu, “Dpm-solver: A fast ode solver for diffusion probabilistic model sampling in around 10 steps,” in Advances in Neural Information Processing Systems (NeurIPS), vol. 35, 2022, pp. 5775–5787

    2022

  22. [30]

    Gans trained by a two time-scale update rule converge to a local nash equilibrium,

    M. Heusel, H. Ramsauer, T. Unterthiner, B. Nessler, and S. Hochreiter, “Gans trained by a two time-scale update rule converge to a local nash equilibrium,” 2018. [Online]. Available: https://arxiv.org/abs/1706.08500

    2018 arXiv

  23. [31]

    Daubechies, Ten lectures on wavelets

    I. Daubechies, Ten lectures on wavelets . SIAM, 1992

    1992

  24. [32]

    Dpm-solver: A fast ode solver for diffusion probabilistic model sampling in around 10 steps,

    C. Lu, Y . Zhou, F. Bao, J. Chen, C. Li, and J. Zhu, “Dpm-solver: A fast ode solver for diffusion probabilistic model sampling in around 10 steps,” Advances in Neural Information Processing Systems , vol. 35, pp. 5775–5787, 2022

    2022

  25. [33]

    Flow straight and fast: Learning to generate and transfer data with rectified flow,

    X. Liu, C. Gong, and Q. Liu, “Flow straight and fast: Learning to generate and transfer data with rectified flow,” arXiv preprint arXiv:2209.03003, 2022. JOURNAL OF LATEX CLASS FILES, VOL. 13, NO. 9, SEPTEMBER 2014 13

    2022 arXiv

  26. [34]

    Residual de- noising diffusion models,

    J. Liu, Q. Wang, H. Fan, Y . Wang, Y . Tang, and L. Qu, “Residual de- noising diffusion models,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2024, pp. 2773–2783

    2024

  27. [35]

    Discrimination among semi-arid landscape endmembers using the spectral angle mapper (sam) algorithm,

    R. H. Yuhas, A. F. Goetz, and J. W. Boardman, “Discrimination among semi-arid landscape endmembers using the spectral angle mapper (sam) algorithm,” in JPL, Summaries of the Third Annual JPL Airborne Geoscience Workshop. Volume 1: AVIRIS Workshop, 1992

    1992

  28. [36]

    Wald, Data fusion: definitions and architectures: fusion of images of different spatial resolutions

    L. Wald, Data fusion: definitions and architectures: fusion of images of different spatial resolutions. Presses des MINES, 2002

    2002

  29. [37]

    Hypercomplex quality assessment of multi/hyperspectral images,

    A. Garzelli and F. Nencini, “Hypercomplex quality assessment of multi/hyperspectral images,” IEEE Geoscience and Remote Sensing Letters, vol. 6, pp. 662–665, 2009

    2009

  30. [38]

    A wavelet transform method to merge landsat tm and spot panchromatic data,

    J. Zhou, D. L. Civco, and J. A. Silander, “A wavelet transform method to merge landsat tm and spot panchromatic data,” International Journal of Remote Sensing , vol. 19, no. 4, pp. 743–757, 1998

    1998

  31. [39]

    A critical comparison among pansharpening algorithms,

    G. Vivone, L. Alparone, J. Chanussot, M. Dalla Mura, A. Garzelli, G. A. Licciardi, R. Restaino, and L. Wald, “A critical comparison among pansharpening algorithms,” IEEE Transactions on Geoscience and Remote Sensing , vol. 53, no. 5, pp. 2565–2586, 2014

    2014

  32. [40]

    Context-driven fusion of high spatial and spectral resolution images based on oversam- pled multiresolution analysis,

    B. Aiazzi, L. Alparone, S. Baronti, and A. Garzelli, “Context-driven fusion of high spatial and spectral resolution images based on oversam- pled multiresolution analysis,” IEEE Transactions on Geoscience and Remote Sensing, vol. 40, no. 10, pp. 2300–2312, 2002

    2002

  33. [41]

    Robust band-dependent spatial-detail approaches for panchromatic sharpening,

    G. Vivone, “Robust band-dependent spatial-detail approaches for panchromatic sharpening,” IEEE Transactions on Geoscience and Re- mote Sensing, vol. 57, no. 9, pp. 6421–6433, 2019

    2019

  34. [42]

    Full scale regression-based injection coefficients for panchromatic sharpening,

    G. Vivone, R. Restaino, and J. Chanussot, “Full scale regression-based injection coefficients for panchromatic sharpening,” IEEE Transactions on Image Processing , vol. 27, no. 7, pp. 3418–3431, 2018

    2018

  35. [43]

    Haze correction for contrast-based multispectral pansharpening,

    S. Lolli, L. Alparone, A. Garzelli, and G. Vivone, “Haze correction for contrast-based multispectral pansharpening,” IEEE Geoscience and Remote Sensing Letters , vol. 14, no. 12, pp. 2255–2259, 2017

    2017

  36. [44]

    Pansharp- ening via detail injection based convolutional neural networks,

    L. He, Y . Rao, J. Li, J. Chanussot, A. Plaza, J. Zhu, and B. Li, “Pansharp- ening via detail injection based convolutional neural networks,” IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, vol. 12, no. 4, pp. 1188–1204, 2019

    2019

  37. [45]

    A multiscale and multidepth convolutional neural network for remote sensing imagery pan-sharpening,

    Q. Yuan, Y . Wei, X. Meng, H. Shen, and L. Zhang, “A multiscale and multidepth convolutional neural network for remote sensing imagery pan-sharpening,” IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing , vol. 11, no. 3, pp. 978–989, 2018

    2018

  38. [46]

    Zero-shot semi- supervised learning for pansharpening,

    Q. Cao, L.-J. Deng, W. Wang, J. Hou, and G. Vivone, “Zero-shot semi- supervised learning for pansharpening,” Information Fusion , vol. 101, p. 102001, 2024

    2024

  39. [47]

    U2net: A general frame- work with spatial-spectral-integrated double u-net for image fusion,

    S. Peng, C. Guo, X. Wu, and L.-J. Deng, “U2net: A general frame- work with spatial-spectral-integrated double u-net for image fusion,” in Proceedings of the 31st ACM International Conference on Multimedia , 2023, pp. 3219–3227

    2023

  40. [48]

    A general paradigm with detail-preserving conditional invertible network for image fusion,

    W. Wang, L.-J. Deng, R. Ran, and G. Vivone, “A general paradigm with detail-preserving conditional invertible network for image fusion,” International Journal of Computer Vision , vol. 132, no. 4, pp. 1029– 1054, 2024

    2024

  41. [49]

    Pan-sharpening with customized transformer and invertible neural network,

    M. Zhou, J. Huang, Y . Fang, X. Fu, and A. Liu, “Pan-sharpening with customized transformer and invertible neural network,” in AAAI Conference on Artificial Intelligence (AAAI) , 2022

    2022

  42. [50]

    Memory-augmented model- driven network for pansharpening,

    K. Yan, M. Zhou, L. Zhang, and C. Xie, “Memory-augmented model- driven network for pansharpening,” in Proceedings of the European Conference on Computer Vision (ECCV) . Springer, 2022, pp. 306– 322

    2022

  43. [51]

    Dynamic cross feature fusion for remote sensing pansharpening,

    X. Wu, T.-Z. Huang, L.-J. Deng, and T.-J. Zhang, “Dynamic cross feature fusion for remote sensing pansharpening,” in Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) , 2021, pp. 14 687–14 696

    2021

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.