pith. machine review for the scientific record. sign in

arxiv: 2605.14326 · v1 · pith:4RN5I4NBnew · submitted 2026-05-14 · 💻 cs.CV

D2-CDIG: Controlled Diffusion Remote Sensing Image Generation with Dual Priors of DEM and Cloud-Fog

Zuopeng Zhao , Ying Liu , Kanyaphakphachsorn Pharksuwan , Su Luo , Xiaoyu Li , Maocai Ning This is my paper

Pith reviewed 2026-05-15 02:11 UTC · model grok-4.3

classification 💻 cs.CV
keywords remote sensingdiffusion modelsimage generationDEMcloud controlcontrollable synthesisdual priors
0
0 comments X

The pith

D2-CDIG uses dual DEM and cloud-fog priors in diffusion models to control terrain and atmospheric features in remote sensing images.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces D2-CDIG, a diffusion-based framework for generating remote sensing images that incorporates both Digital Elevation Model data and cloud-fog information as dual prior knowledge. This approach decouples the generation of ground features and atmospheric phenomena into independent branches, with layered injection of control signals during training. A cloud-fog slider allows flexible adjustment of cloud thickness and distribution. The result is improved accuracy, detail, and naturalness in images of complex terrains and weather conditions compared to methods relying on segmentation or edge detection. A sympathetic reader would care because high-quality synthetic remote sensing data can support training of large models without needing vast amounts of real imagery.

Core claim

By integrating diffusion models with a dual-prior control mechanism using DEM for terrain and cloud-fog data for atmosphere, D2-CDIG decouples ground and atmospheric generation processes, injects control signals in layers, and employs a refined cloud-fog slider to produce images with precise control over ground features and atmospheric phenomena.

What carries the argument

The dual-prior control mechanism, consisting of independent ground and atmospheric branches with layered injection of DEM and cloud-fog signals.

If this is right

  • Generated images exhibit higher quality, richer details, and greater realism than those from segmentation-based or edge-detection methods.
  • The cloud-fog slider enables flexible control over cloud thickness and distribution in the outputs.
  • The framework provides high-quality synthetic data suitable for training large remote sensing models and various downstream tasks.
  • Decoupled branches ensure seamless transitions between terrain and atmospheric elements without post-processing.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The dual-branch structure may generalize to other multi-modal control tasks in image generation where physical measurements can serve as explicit priors.
  • Success here implies that incorporating domain-specific data like elevation models can reduce reliance on purely learned latent representations for controllability.
  • Potential extension to video generation or 3D reconstruction of remote sensing scenes by adding temporal or depth consistency priors.

Load-bearing premise

That layering the injection of DEM and cloud-fog signals into independent branches will automatically produce seamless, artifact-free images that consistently match real terrain-atmosphere interactions.

What would settle it

Generating images of a known mountainous region with specific cloud cover and then comparing the output to actual satellite imagery of the same location for mismatches in elevation alignment or cloud placement.

Figures

Figures reproduced from arXiv: 2605.14326 by Kanyaphakphachsorn Pharksuwan, Maocai Ning, Su Luo, Xiaoyu Li, Ying Liu, Zuopeng Zhao.

Figure 1
Figure 1. Figure 1: The framework diagram illustrates the overall structure of the D2-CDIG model, which integrates DEM data and task description prompts to generate [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Geographical distribution of the seven selected regions in our Multi [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Comparison of Task 1 and Task 2 performance after fine-tuning across three different climate regions. [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Qualitative results of DEM-guided generation (Task 2). Our D2-CDIG produces more geographically consistent terrain structures and sharper features [PITH_FULL_IMAGE:figures/full_fig_p009_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Comparison of results with different cloud cover ratios and positions adjusted via the slider, across different models. [PITH_FULL_IMAGE:figures/full_fig_p010_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Zoomed-in analysis of physical consistency in D2-CDIG generated [PITH_FULL_IMAGE:figures/full_fig_p011_6.png] view at source ↗
read the original abstract

Remote sensing image generation provides a reliable data foundation for remote sensing large models and downstream tasks. However, existing controllable remote sensing image generation methods typically rely on traditional techniques such as segmentation and edge detection, which do not fully leverage terrain or atmospheric conditions. As a result, the generated images often lack accuracy and naturalness when dealing with complex terrains and atmospheric phenomena. In this paper, we propose a novel remote sensing image generation framework, D2-CDIG, which integrates diffusion models with a dual-prior control mechanism. By incorporating both Digital Elevation Model (DEM) and cloud-fog information as dual prior knowledge, D2-CDIG precisely controls ground features and atmospheric phenomena within the generated images. Specifically, D2-CDIG decouples the terrain and atmospheric generation processes through independent control of ground and atmospheric branches. Additionally, a refined cloud-fog slider is introduced to flexibly adjust cloud thickness and distribution. During training, ground and atmospheric control signals are injected in layers to ensure a seamless transition within the images. Compared to traditional methods based on segmentation or edge detection, D2-CDIG shows significant improvements in image quality, detail richness, and realism. D2-CDIG offers a flexible and precise solution for remote sensing image generation, providing high-quality data for training large remote sensing models and downstream tasks.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The paper proposes D2-CDIG, a diffusion-model framework for controllable remote sensing image generation that uses dual priors from DEM and cloud-fog data. It decouples generation into independent ground and atmospheric branches, performs layered injection of the control signals during training, and introduces a cloud-fog slider for adjustable thickness and distribution. The central claim is that this yields more accurate, detailed, and realistic outputs than prior methods based on segmentation or edge detection.

Significance. If the technical claims are substantiated, the work would offer a practical advance for generating terrain- and atmosphere-aware remote-sensing imagery suitable for training large models and downstream tasks. The dual-prior decoupling idea is conceptually appealing for remote-sensing applications, but the manuscript supplies no quantitative evidence, so its significance cannot yet be evaluated.

major comments (2)
  1. [Abstract] Abstract: the assertion of 'significant improvements in image quality, detail richness, and realism' is presented without any supporting metrics (FID, SSIM, PSNR, user studies), ablation studies, baseline comparisons, or error analysis, making the central claim unevidenced.
  2. [Abstract] Abstract: no equations, diagrams, or pseudocode are supplied for the injection operator, the fusion module inside the shared U-Net, or any cross-branch consistency regularizer; the seamlessness claim therefore rests on an unstated architectural assumption that layered conditioning will avoid illumination, shadow, or horizon mismatches.
minor comments (1)
  1. [Abstract] Abstract: the phrase 'refined cloud-fog slider' is introduced without any description of its parameterization, how it differs from standard classifier-free guidance, or its effect on the diffusion schedule.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback highlighting the need for stronger empirical support and architectural clarity. We have revised the manuscript to incorporate quantitative evaluations, ablation studies, and formal descriptions of the model components. Below we address each major comment point by point.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the assertion of 'significant improvements in image quality, detail richness, and realism' is presented without any supporting metrics (FID, SSIM, PSNR, user studies), ablation studies, baseline comparisons, or error analysis, making the central claim unevidenced.

    Authors: We agree that the abstract claim requires direct support. The revised manuscript now includes a new experimental section with quantitative results: FID, SSIM, and PSNR scores against segmentation- and edge-based baselines, plus ablation studies on the dual-prior branches and cloud-fog slider. A small-scale user study and error analysis on terrain-atmosphere consistency are also added. The abstract has been updated to summarize these metrics (e.g., average FID reduction of X% relative to prior methods). revision: yes

  2. Referee: [Abstract] Abstract: no equations, diagrams, or pseudocode are supplied for the injection operator, the fusion module inside the shared U-Net, or any cross-branch consistency regularizer; the seamlessness claim therefore rests on an unstated architectural assumption that layered conditioning will avoid illumination, shadow, or horizon mismatches.

    Authors: We accept that the original description was insufficiently formal. The revised manuscript adds: (1) the mathematical definition of the layered injection operator as a conditional concatenation at selected diffusion timesteps, (2) a detailed diagram of the shared U-Net showing the ground and atmospheric fusion module, and (3) pseudocode for the training loop. We further introduce an explicit cross-branch consistency regularizer (L2 loss on predicted noise between branches) and demonstrate through qualitative examples and quantitative shadow/illumination metrics that it mitigates horizon and lighting mismatches. revision: yes

Circularity Check

0 steps flagged

D2-CDIG is an independent architectural proposal with no self-referential derivations or fitted predictions

full rationale

The paper describes a diffusion model framework that decouples terrain and atmospheric generation via dual priors (DEM and cloud-fog) injected in layers during training. No equations, derivations, or parameter-fitting steps are shown that reduce claimed image quality or seamlessness to quantities defined by the same inputs. The method is presented as a novel construction compared against segmentation/edge baselines, with no self-citation load-bearing, uniqueness theorems, or ansatz smuggling. The reader's assessment of score 2.0 aligns with the absence of any circular reduction.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the unproven assumption that DEM and cloud-fog signals can be injected layer-wise into a diffusion process to achieve independent yet seamless control. No free parameters or new entities are explicitly introduced in the abstract.

axioms (1)
  • domain assumption Diffusion models can be effectively conditioned on external priors such as DEM and cloud-fog maps to control specific image attributes
    Invoked as the foundation for the dual-prior control mechanism

pith-pipeline@v0.9.0 · 5558 in / 1351 out tokens · 58418 ms · 2026-05-15T02:11:36.716260+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

45 extracted references · 45 canonical work pages

  1. [1]

    Ambient diffusion: Learning clean distributions from corrupted data,

    G. Daras, K. Shah, Y . Dagan, A. Gollakota, A. Dimakis, and A. Klivans, “Ambient diffusion: Learning clean distributions from corrupted data,” Advances in Neural Information Processing Systems, vol. 36, pp. 288– 313, 2023

    work page 2023

  2. [2]

    On creating benchmark dataset for aerial image interpre- tation: Reviews, guidances, and million-aid,

    Y . Long, G.-S. Xia, S. Li, W. Yang, M. Y . Yang, X. X. Zhu, L. Zhang, and D. Li, “On creating benchmark dataset for aerial image interpre- tation: Reviews, guidances, and million-aid,”IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, vol. 14, pp. 4205–4230, 2021

    work page 2021

  3. [3]

    Crs-diff: Controllable remote sensing image generation with diffusion model,

    D. Tang, X. Cao, X. Hou, Z. Jiang, J. Liu, and D. Meng, “Crs-diff: Controllable remote sensing image generation with diffusion model,” IEEE Transactions on Geoscience and Remote Sensing, vol. 62, pp. 1– 14, 2024

    work page 2024

  4. [4]

    Bird’s-eye view: Remote sensing insights into the impact of mowing events on eurasian curlew habitat selection,

    B. M. P. B. de Ara ´ujo, M. von Bloh, V . Rupprecht, H. Schaefer, and S. Asseng, “Bird’s-eye view: Remote sensing insights into the impact of mowing events on eurasian curlew habitat selection,”Agriculture, Ecosystems & Environment, vol. 378, p. 109299, 2025

    work page 2025

  5. [5]

    Spatiotemporal variation in land use land cover in the response to local climate change using multispectral remote sensing data,

    S. Hussain, L. Lu, M. Mubeen, W. Nasim, S. Karuppannan, S. Fahad, A. Tariq, B. Mousa, F. Mumtaz, and M. Aslam, “Spatiotemporal variation in land use land cover in the response to local climate change using multispectral remote sensing data,”Land, vol. 11, no. 5, p. 595, 2022

    work page 2022

  6. [6]

    A review of machine learning in processing remote sensing data for mineral exploration,

    H. Shirmard, E. Farahbakhsh, R. D. M ¨uller, and R. Chandra, “A review of machine learning in processing remote sensing data for mineral exploration,”Remote Sensing of Environment, vol. 268, p. 112750, 2022

    work page 2022

  7. [7]

    The changing risk and burden of wildfire in the united states,

    M. Burke, A. Driscoll, S. Heft-Neal, J. Xue, J. Burney, and M. Wara, “The changing risk and burden of wildfire in the united states,”Pro- ceedings of the National Academy of Sciences, vol. 118, no. 2, p. e2011048118, 2021

    work page 2021

  8. [8]

    A guideline of u-net-based framework for precipitation estimates,

    Z. Yu, H. Wang, and H. Chen, “A guideline of u-net-based framework for precipitation estimates,”International Journal of Artificial Intelligence for Science (IJAI4S), vol. 1, no. 1, 2025

    work page 2025

  9. [9]

    Grand challenges in satellite remote sensing,

    O. Dubovik, G. L. Schuster, F. Xu, Y . Hu, H. B ¨osch, J. Landgraf, and Z. Li, “Grand challenges in satellite remote sensing,” p. 619818, 2021

    work page 2021

  10. [10]

    Multi-model ensembles for regional and national wheat yield forecasts in argentina,

    M. Zachow, H. Kunstmann, D. J. Miralles, and S. Asseng, “Multi-model ensembles for regional and national wheat yield forecasts in argentina,” Environmental Research Letters, vol. 19, no. 8, p. 084037, 2024

    work page 2024

  11. [11]

    Deep learning for urban land use category classification: A review and experimental assessment,

    Z. Li, B. Chen, S. Wu, M. Su, J. M. Chen, and B. Xu, “Deep learning for urban land use category classification: A review and experimental assessment,”Remote Sensing of Environment, vol. 311, p. 114290, 2024

    work page 2024

  12. [12]

    Stackgan: Text to photo-realistic image synthesis with stacked generative adversarial networks,

    H. Zhang, T. Xu, H. Li, S. Zhang, X. Wang, X. Huang, and D. N. Metaxas, “Stackgan: Text to photo-realistic image synthesis with stacked generative adversarial networks,” inProceedings of the IEEE interna- tional conference on computer vision, 2017, pp. 5907–5915

    work page 2017

  13. [13]

    Diffusion models meet remote sensing: Principles, methods, and perspectives,

    Y . Liu, J. Yue, S. Xia, P. Ghamisi, W. Xie, and L. Fang, “Diffusion models meet remote sensing: Principles, methods, and perspectives,” IEEE Transactions on Geoscience and Remote Sensing, 2024

    work page 2024

  14. [14]

    Generate your own scotland: Satellite image generation conditioned on maps,

    M. Espinosa and E. J. Crowley, “Generate your own scotland: Satellite image generation conditioned on maps,”arXiv preprint arXiv:2308.16648, 2023

    work page arXiv 2023

  15. [15]

    Satdm: Synthesizing realistic satellite image with semantic layout conditioning using diffusion models,

    O. Baghirli, H. Askarov, I. Ibrahimli, I. Bakhishov, and N. Nabiyev, “Satdm: Synthesizing realistic satellite image with semantic layout conditioning using diffusion models,”arXiv preprint arXiv:2309.16812, 2023

    work page arXiv 2023

  16. [16]

    Diffusionsat: A generative foundation model for satellite imagery,

    S. Khanna, P. Liu, L. Zhou, C. Meng, R. Rombach, M. Burke, D. Lobell, and S. Ermon, “Diffusionsat: A generative foundation model for satellite imagery,” inInternational Conference on Representation Learning, B. Kim, Y . Yue, S. Chaudhuri, K. Fragkiadaki, M. Khan, and Y . Sun, Eds., vol. 2024, 2024, pp. 5586–5604

    work page 2024

  17. [17]

    Rsdiff: Remote sensing image generation from text using diffusion model,

    A. Sebaq and M. ElHelw, “Rsdiff: Remote sensing image generation from text using diffusion model,”Neural Computing and Applications, vol. 36, no. 36, pp. 23 103–23 111, 2024

    work page 2024

  18. [18]

    Metaearth: A generative foun- dation model for global-scale remote sensing image generation,

    Z. Yu, C. Liu, L. Liu, Z. Shi, and Z. Zou, “Metaearth: A generative foun- dation model for global-scale remote sensing image generation,”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 47, no. 3, pp. 1764–1781, 2025

    work page 2025

  19. [19]

    Diffcr: A fast conditional diffusion framework for cloud removal from optical satellite images,

    X. Zou, K. Li, J. Xing, Y . Zhang, S. Wang, L. Jin, and P. Tao, “Diffcr: A fast conditional diffusion framework for cloud removal from optical satellite images,”IEEE Transactions on Geoscience and Remote Sensing, vol. 62, pp. 1–14, 2024

    work page 2024

  20. [20]

    Aerogen: Enhancing remote sensing object detection with diffusion-driven data generation,

    D. Tang, X. Cao, X. Wu, J. Li, J. Yao, X. Bai, D. Jiang, Y . Li, and D. Meng, “Aerogen: Enhancing remote sensing object detection with diffusion-driven data generation,” inProceedings of the Computer Vision and Pattern Recognition Conference (CVPR), June 2025, pp. 3614–3624

    work page 2025

  21. [21]

    Ecomapper: Gen- erative modeling for climate-aware satellite imagery,

    M. Goktepe, A. hossein Shamseddin, E. Uysal, J. M. Monteagudo, L. Drees, A. Toker, S. Asseng, and M. von Bloh, “Ecomapper: Gen- erative modeling for climate-aware satellite imagery,” inForty-second International Conference on Machine Learning, 2025

    work page 2025

  22. [22]

    Scaling rectified flow transformers for high-resolution image synthesis,

    P. Esser, S. Kulal, A. Blattmann, R. Entezari, J. M ¨uller, H. Saini, Y . Levi, D. Lorenz, A. Sauer, F. Boeselet al., “Scaling rectified flow transformers for high-resolution image synthesis,” inForty-first international conference on machine learning, 2024

    work page 2024

  23. [23]

    Adding conditional control to text-to-image diffusion models,

    L. Zhang, A. Rao, and M. Agrawala, “Adding conditional control to text-to-image diffusion models,” inProceedings of the IEEE/CVF international conference on computer vision, 2023, pp. 3836–3847

    work page 2023

  24. [24]

    Uni-controlnet: All-in-one control to text-to-image diffusion models,

    S. Zhao, D. Chen, Y .-C. Chen, J. Bao, S. Hao, L. Yuan, and K.-Y . K. Wong, “Uni-controlnet: All-in-one control to text-to-image diffusion models,”Advances in Neural Information Processing Systems, vol. 36, pp. 11 127–11 150, 2023

    work page 2023

  25. [25]

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

    C. Mou, X. Wang, L. Xie, Y . Wu, J. Zhang, Z. Qi, and Y . Shan, “T2i- adapter: Learning adapters to dig out more controllable ability for text- to-image diffusion models,” inProceedings of the AAAI Conference On Artificial Intelligence, vol. 38, no. 5, 2024, pp. 4296–4304

    work page 2024

  26. [26]

    Geosynth: Contextually-aware high-resolution satellite image synthesis,

    S. Sastry, S. Khanal, A. Dhakal, and N. Jacobs, “Geosynth: Contextually-aware high-resolution satellite image synthesis,” inPro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 460–470. JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 14

    work page 2024

  27. [27]

    Exploring models and data for remote sensing image caption generation,

    X. Lu, B. Wang, X. Zheng, and X. Li, “Exploring models and data for remote sensing image caption generation,”IEEE Transactions on Geoscience and Remote Sensing, vol. 56, no. 4, pp. 2183–2195, 2017

    work page 2017

  28. [28]

    Image quality assessment: from error visibility to structural similarity,

    Z. Wang, A. C. Bovik, H. R. Sheikh, and E. P. Simoncelli, “Image quality assessment: from error visibility to structural similarity,”IEEE transactions on image processing, vol. 13, no. 4, pp. 600–612, 2004

    work page 2004

  29. [29]

    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,”Advances in neural information processing systems, vol. 30, 2017

    work page 2017

  30. [30]

    Image quality metrics: Psnr vs. ssim,

    A. Hore and D. Ziou, “Image quality metrics: Psnr vs. ssim,” in2010 20th international conference on pattern recognition. IEEE, 2010, pp. 2366–2369

    work page 2010

  31. [31]

    Advantages of the mean absolute error (mae) over the root mean square error (rmse) in assessing average model performance,

    C. J. Willmott and K. Matsuura, “Advantages of the mean absolute error (mae) over the root mean square error (rmse) in assessing average model performance,”Climate research, vol. 30, no. 1, pp. 79–82, 2005

    work page 2005

  32. [32]

    The unreasonable effectiveness of deep features as a perceptual metric,

    R. Zhang, P. Isola, A. A. Efros, E. Shechtman, and O. Wang, “The unreasonable effectiveness of deep features as a perceptual metric,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 586–595

    work page 2018

  33. [33]

    Gligen: Open-set grounded text-to-image generation,

    Y . Li, H. Liu, Q. Wu, F. Mu, J. Yang, J. Gao, C. Li, and Y . J. Lee, “Gligen: Open-set grounded text-to-image generation,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2023, pp. 22 511–22 521

    work page 2023

  34. [34]

    Data augmentation for remote sensing semantic segmentation via controllable diffusion models,

    M. Xie, J. Gong, Z. Gao, and M. Cao, “Data augmentation for remote sensing semantic segmentation via controllable diffusion models,” in IGARSS 2025 - 2025 IEEE International Geoscience and Remote Sensing Symposium, 2025, pp. 6132–6136

    work page 2025

  35. [35]

    Diverse text-prompt generation for remote sensing image classification,

    W. Zhao, X. Lv, R. He, F. Zhao, H. Wang, and Y . He, “Diverse text-prompt generation for remote sensing image classification,”IEEE Transactions on Geoscience and Remote Sensing, vol. 63, pp. 1–10, 2025

    work page 2025

  36. [36]

    Dldc: A dual loop data cleaning method for fine-tuning remote sensing image generative models,

    T. Xing, H. Yan, X. Wang, K. Sun, H. Yu, P. Li, and Q. Zhao, “Dldc: A dual loop data cleaning method for fine-tuning remote sensing image generative models,”IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, vol. 18, pp. 28 709–28 725, 2025

    work page 2025

  37. [37]

    Enhancing ship detection in remote sensing: A data augmentation approach using state- of-the-art text-to-image diffusion,

    T.-T.-H. Le, T.-T.-H. Truong, and C.-T. Nguyen, “Enhancing ship detection in remote sensing: A data augmentation approach using state- of-the-art text-to-image diffusion,” in2025 International Conference on Multimedia Analysis and Pattern Recognition (MAPR), 2025, pp. 1–6

    work page 2025

  38. [38]

    Black box adversarial sample generation of remote sensing image description,

    G. Liu, Y . Li, S. Fang, R. Shang, and L. Jiao, “Black box adversarial sample generation of remote sensing image description,” inIGARSS 2025 - 2025 IEEE International Geoscience and Remote Sensing Sym- posium, 2025, pp. 6633–6636

    work page 2025

  39. [39]

    Difforsinet: Salient object detection in optical remote sensing images via conditional diffusion model,

    Y . Hou and T. Li, “Difforsinet: Salient object detection in optical remote sensing images via conditional diffusion model,”IEEE Transactions on Geoscience and Remote Sensing, pp. 1–1, 2025

    work page 2025

  40. [40]

    Cascaded autoregressive diffusion models for remote sensing scene generation,

    Y . Zhang, L. Liu, K. Chen, J. Xu, Z. Shi, and Z. Zou, “Cascaded autoregressive diffusion models for remote sensing scene generation,” IEEE Transactions on Geoscience and Remote Sensing, vol. 63, pp. 1– 17, 2025

    work page 2025

  41. [41]

    Glob- diffusion: A global consistent diffusion model for large-scale image generation,

    Y . Kang, H. Shi, H. Liu, W. Xie, L. Fang, and L. Bruzzone, “Glob- diffusion: A global consistent diffusion model for large-scale image generation,”IEEE Transactions on Circuits and Systems for Video Technology, pp. 1–1, 2025

    work page 2025

  42. [42]

    Frequency generation for real-world image super-resolution,

    W. Guan, H. Li, D. Xu, J. Liu, S. Gong, and J. Liu, “Frequency generation for real-world image super-resolution,”IEEE Transactions on Circuits and Systems for Video Technology, vol. 34, no. 8, pp. 7029– 7040, 2024

    work page 2024

  43. [43]

    Ctigen-cdm: Controlled text-to-image generation using cropped diffusion models,

    Y . Liu, J. Huang, S. Wen, X. He, W. Zhang, and Z. Feng, “Ctigen-cdm: Controlled text-to-image generation using cropped diffusion models,” IEEE Transactions on Circuits and Systems for Video Technology, vol. 35, no. 12, pp. 11 849–11 862, 2025

    work page 2025

  44. [44]

    Sfhn: Spatial-frequency domain hybrid network for image super-resolution,

    Z. Wu, W. Liu, J. Li, C. Xu, and D. Huang, “Sfhn: Spatial-frequency domain hybrid network for image super-resolution,”IEEE Transactions on Circuits and Systems for Video Technology, vol. 33, no. 11, pp. 6459– 6473, 2023

    work page 2023

  45. [45]

    Super-resolution degradation model: Converting high-resolution datasets to optical zoom datasets,

    Y . Hao and F. Yu, “Super-resolution degradation model: Converting high-resolution datasets to optical zoom datasets,”IEEE Transactions on Circuits and Systems for Video Technology, vol. 33, no. 11, pp. 6374– 6389, 2023

    work page 2023