Pith. sign in

REVIEW 4 major objections 5 minor 47 references

Hyperspectral Image Generation with Unmixing Guided Diffusion Model

T0 review · 4 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read A diffusion model that generates hyperspectral images in the physically constrained abundance space achieves both high fidelity and diversity across five benchmark scenes.

desk verdict The architecture is a plausible incremental step, but the evaluation is built on a circular fidelity metric and ad hoc ratio, so the SOTA claim does not hold. read the letter →

arxiv 2506.02601 v3 pith:4YLV7GUV submitted 2025-06-03 cs.CV eess.IV

Shiyu Shen , Bin Pan , Ziye Zhang , Zhenwei Shi This is my paper
classification cs.CVeess.IV
keywords hyperspectralimagegenerationunmixingdiffusionmodelabundancespacelinearmixingpointfidelityblockdiversityunconditional
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 an unconditional generative model for hyperspectral images can be made both physically valid and diverse by moving generation out of the high-dimensional spectral space and into the low-dimensional abundance space of the linear mixing model. The proposed HUD model combines a frozen unmixing autoencoder, initialized from extracted endmembers, with a diffusion process that runs on a constraint-relaxed version of the abundance maps and maps back through a softmax, so every generated pixel obeys non-negativity and sum-to-one. This lets the generator synthesize full images whose spatial layout is not conditioned on any input, unlike prior spectral-super-resolution or segmentation-guided methods. The paper also introduces two hyperspectral-specific metrics, point fidelity and block diversity, and reports that HUD attains the highest fidelity and the best combined fidelity-diversity ratio across five benchmark scenes. Such a generator could supply scarce training data for downstream remote sensing tasks without requiring paired RGB images or labels.

What carries the argument

The load-bearing mechanism is the unmixing autoencoder plus a differentiable, near-lossless constraint-relaxing projection. The linear mixing model $Y = AX + \varepsilon$ is encoded by $X = (A^T A)^{-1}A^T Y$ and decoded as $\hat{Y} = A\hat{X}$, where $A$ contains $d$ endmembers extracted by an unmixing algorithm such as VCA. The diffusion model never sees raw spectra; it sees $Z = \ln(X + e^{-\ln d - 8})$ inside the unconstrained domain and converts back with $\hat{X}[i,j,k] = e^{Z[i,j,k]} / \sum_i e^{Z[i,j,k]}$, so non-negativity and sum-to-one hold by construction. This reduces a hundred-band generation problem to generation over $d$ abundance maps, with $d$ set to the number of classes in each benchmark.

What would settle it

Run the pipeline on a scene whose true number of distinct materials is known to differ from the labeled class count, or on a scene with known nonlinear mixing; if the frozen unmixing autoencoder's reconstruction error on held-out pixels is large, or if the generated spectra fail the paper's own point-fidelity test against real spectra, then the assumed abundance space is not faithful and the method's central premise is refuted.

Watch

Extended reading notes

Core claim

On its own terms, the central discovery is that the abundance manifold defined by linear hyperspectral unmixing is a viable and beneficial latent space for diffusion-based hyperspectral image synthesis. The autoencoder is built directly from the physics: the decoder is initialized as the endmember matrix $A$ and the encoder as its pseudo-inverse $(A^T A)^{-1}A^T$, and because both are frozen the model cannot drift away from the linear mixing model. The diffusion module treats the projected variable $Z_{i,j,k} = \ln(X_{i,j,k} + e^{-\ln d -8})$ as an unconstrained Gaussian-like target and reconstructs abundances by a softmax, which is shown to be a near-lossless transformation for typical abundance magnitudes. Empirically, HUD reports the highest point fidelity on all five datasets and the lowest ratio of block diversity to point fidelity, indicating that generated images preserve spectral content and are not just copies of the training scene.

Load-bearing premise

The main load-bearing premise is that the number of materials in a scene equals the number of ground-truth classes used to set the abundance dimension, and that a linear mixing model with one endmember matrix can faithfully represent every pixel; if the true scene has more materials, nonlinear mixing, or spectral variability, the abundance space cannot reconstruct valid spectra and the generated images inherit that error.

Editorial extensions

If this is right

  • If HUD works as claimed, hyperspectral data augmentation no longer needs paired RGB images or segmentation maps, so synthetic scenes can introduce spatial arrangements absent from the original data.
  • Generating in the abundance space cuts the diffusion model's output dimensionality from hundreds of spectral bands to a small number of endmember abundance maps, which could make larger spatial footprints tractable.
  • Because the softmax reconstruction enforces the physical constraints by construction, every sampled image lies inside the linear mixing model, reducing the chance of physically impossible spectra.
  • The proposed point-fidelity and block-diversity metrics offer a way to evaluate unconditional HSI generation without ImageNet-pretrained classifiers, which are unreliable for small hyperspectral datasets.
  • The method's ability to produce new roads and buildings in the generated scenes suggests spatial diversity beyond direct copying, a property that prior conditional methods do not have.

Reading between the lines

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

  • A natural extension is to replace the class-count choice of $d$ with an automatic endmember-number estimate; then the same pipeline could generate scenes whose material composition was not assumed in advance, at the cost of evaluating how the softmax link interacts with the estimated endmembers.
  • Because the projection pair is the logistic-normal (softmax) link, the framework can be read as latent diffusion on a compositional data manifold; one could test whether a Gaussian diffusion on the log-ratio transform and a Dirichlet-flow diffusion give different fidelity-diversity trade-offs.
  • The point fidelity metric rewards each generated pixel for matching the closest real pixel, so a model that reproduces common spectra but never explores rare ones could still score highly; a useful stress test is to count how many distinct abundance modes the generator actually visits versus the number of endmembers.
  • The frozen autoencoder is both the strength and the limiting factor: if spectral variability within a scene is large, a single endmember matrix cannot represent all pixels, and fine-tuning the encoder on abundant multi-scene data would be needed to keep the generative manifold faithful.
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 HUD, a hyperspectral image generation framework that couples a VCA-initialized, frozen unmixing autoencoder with a diffusion model operating on a log-abundance space. A differentiable softmax projection (Eqs. 4-5) maps the unconstrained diffusion output back to the non-negative, sum-to-one abundance simplex, and the final image is reconstructed as Y_hat = A X_hat. The authors introduce two evaluation metrics, point fidelity Fp (Eq. 16) and block diversity Db (Eq. 17), and report experiments on five benchmark datasets against GAN, VAE, MPRNet, UD, and UBF, claiming state-of-the-art quality and diversity.

Significance. The core idea of generating hyperspectral images via a physically constrained abundance manifold is well motivated and the DDPM machinery is standard and correctly presented; the projection identity in Eq. (6) is also correct. If the empirical claims held, the work would be a useful contribution to unconditional HSI synthesis. However, the reported evaluation does not support the central claims: the point-fidelity metric is near-saturated by construction, the diversity numbers in the paper's own Table I contradict the diversity claim, and the experimental protocol lacks held-out data and statistical error bars. These issues are load-bearing because the abstract and conclusion rest on them.

major comments (4)
  1. [IV-C1, Eq. (16), Algorithm 2] The point fidelity metric Fp cannot validate the fidelity claim because it is structurally biased by the method's own decoder. In Algorithm 2, every generated pixel is \hat y = A \hat x, where A is the VCA endmember matrix extracted from the real image and \hat x lies on the simplex. Under the paper's own linear mixing model (Eq. 1), real pixels approximately lie in the same simplex spanned by A. The cosine similarity of a generated pixel to its nearest real pixel is therefore high almost by construction, regardless of whether the generative model has learned the distribution. This is confirmed by Table I, where GAN, MPRNet, and HUD all report Fp between 0.995 and 0.999; the metric saturates and cannot separate a good generative model from a trivial decoder that recombines endmembers. The claim that HUD 'achieves the highest point fidelity' is therefore an artifact of the metric rather than evidence of superior generation quality.
  2. [Table I, Section IV-C1] The paper's diversity claim is directly contradicted by its own block-diversity numbers. Db is defined so that lower values indicate more diversity. In Table I, HUD has consistently higher (worse) Db than UD, UBF, and VAE on every dataset: e.g., Indianpines HUD 0.978 vs UD 0.553; KSC HUD 0.990 vs UD 0.833; Pavia HUD 0.912 vs UD 0.818; PaviaU HUD 0.919 vs UD 0.823; Salinas HUD 0.981 vs UD 0.545. HUD ranks first only on the composite Db/Fp ratio, whose numerator is dominated by the saturated Fp values. Thus the abstract's statement that HUD produces HSIs with 'high quality and diversity' is not supported; at best the experiments show that HUD decodes through the VCA simplex, not that it generates diverse spatial-spectral content.
  3. [IV-A, IV-C] The evaluation protocol has no held-out data and no statistical reliability measures. The models are trained on randomly cropped 32x32 sub-images from each original image, and the generated images are then compared against the same original image in the fidelity and diversity metrics. This means the metrics are computed on the training distribution and do not assess generalization to unseen scenes or robustness to overfitting. Furthermore, no error bars or repeated-seed results are reported for any metric, so it is impossible to determine whether the differences in Tables I-III are significant. To support the SOTA claim, the authors need a held-out evaluation protocol and multiple independent runs with reported variance.
  4. [IV-A, Eq. (2)] The latent dimension d of the abundance space is set to the number of ground-truth classes in each dataset, with no validation that the class count equals the intrinsic number of endmembers. If the true endmember dimensionality differs, the VCA endmember matrix A will span the wrong subspace, and the diffusion model will generate abundances that cannot reconstruct physically valid spectra. The paper should justify or estimate d, and should ablate its choice. This is a load-bearing assumption because the entire generative process, and the near-saturated Fp values, depend on the appropriateness of A extracted using this d.
minor comments (5)
  1. [Eq. (17)] In Eq. (17), the inner maximization is written as max over y_b in X, but X denotes the abundance matrix in the paper's notation; it should be Y (the real HSI blocks).
  2. [Algorithm 2] The sampling update formula is missing parentheses around the noise term and does not specify how sigma_t is chosen after the simplification in Eq. (13); this makes the pseudocode hard to reproduce.
  3. [Section IV-C1] The introductory sentence of Section IV-C1 refers to 'region diversity' while the proposed metric and Table I use 'block diversity'; please use consistent terminology.
  4. [References] There are numerous typos in the reference list, e.g., 'Umixing' for 'Unmixing' in Refs. [15], [34], [35], and 'Transanctions' in the biography of Zhenwei Shi; the manuscript would benefit from a careful proofread.
  5. [Section III-C, Eq. (6)] The text calls the projection pair 'near-lossless for typical magnitudes,' but Eq. (6) shows a constant additive and multiplicative shift of the entire abundance vector; it may be worth clarifying the relative error and its dependence on the number of endmembers d.

Circularity Check

2 steps flagged · score 6.0 of 10

HUD's SOTA claim rests on a point-fidelity metric that is inflated by construction: the decoder is the VCA endmember matrix of the same real image used as the Fp reference, while the paper's own Db numbers contradict the diversity claim.

  1. self definitional [Section III-B, Section III-D Eq. (16), Algorithm 2, Table I]
    "we initialize a linear layer as the decoder using the endmember matrix A ... Algorithm 2: ˆX = softmax(Z0); ˆY = D( ˆX) ... Fp = 1 | ˆY | X ˆy∈ ˆY max y∈Y ..."

    Every generated pixel is \ hat y = A \ hat x with \ hat x on the probability simplex (Eq. 5), and A is the VCA endmember matrix extracted from the same real image Y used as the Fp reference. Under the paper's own linear mixing model Y = AX + ε (Eq. 1), real pixels approximately occupy the same endmember simplex, so the cosine similarity of any generated pixel to some real pixel is high almost regardless of diffusion output. Fp therefore measures, by construction, how well the generator stays inside the reference image's endmember hull, not independent sample quality. Table I's saturation (GAN/MPRNet 0.995-0.999) confirms the metric cannot separate a good generative model from a trivial endmember recombiner; the claimed 'highest point fidelity' is forced by the frozen VCA decoder.

  2. other [Section IV-C1 and Table I]
    "UD, UBF and V AE generates the most diverse images, but their poor point fidelity result in overall low-quality generations. ... Combining both point and block diversity, HUD delivers the best overall performance, followed by GAN."

    Table I shows HUD's Db values (0.912-0.990 across datasets) are worse (larger) than VAE, UD, and UBF on every dataset—e.g., Indianpines 0.978 vs UD 0.553—so the 'diversity' half of the headline claim is contradicted by the paper's own numbers. The 'best overall' conclusion is obtained by dividing Db by the construction-inflated Fp, producing the Db/Fp ratio. Since Fp is saturated by the VCA-decoder mechanism, the ratio is minimized by any method that decodes through the same endmember matrix; it is an ad hoc composite that inherits the Fp bias and provides no independent support for the diversity or SOTA claim.

full rationale

The generative pipeline itself—encoding to abundances, diffusing in a constraint-relaxed logit space, and decoding through a fixed endmember matrix—is internally consistent and is not derivational circularity: the diffusion model is trained on VCA-initialized abundances, no parameter is fitted to the Fp target, and the linear mixing model plus VCA are independent prior results. No load-bearing self-citation chain is present. The circularity is concentrated in the headline evaluation. Fp (Eq. 16) uses the real HSI as its reference while Algorithm 2 reconstructs every sample as A\ hat X, where A is the VCA endmember matrix of that same real HSI; under the paper's own model Y ≈ AX, every simplex abundance maps into the real image's endmember hull, so max-cosine Fp is high by construction. This is confirmed by GAN and MPRNet also reaching 0.995-0.999. The paper then leans on Fp and the ad hoc Db/Fp ratio to declare best overall performance and 'high quality and diversity,' even though Table I shows HUD's Db is consistently worse (larger) than VAE/UD/UBF. The IS/FID results are mixed and the paper itself calls them unstable and questionable. Thus the central SOTA claim partially reduces to a construction-inflated metric, while the method retains independent content—hence a partial-circularity score of 6.

Assumptions & free parameters 2 free parameters · 3 assumptions · 2 invented entities

The central claim rests on the linear mixing model, the reliability of VCA endmembers, and a label-derived endmember count. The proposed evaluation metrics are not independently validated and are partially aligned with the model's construction.

free parameters (2)
  • number of endmembers d = equals the number of classes per dataset (16 for Indianpines, 13 for KSC, 9 for Pavia/PaviaU, 16 for Salinas)
    The dimensionality of the abundance manifold is chosen as the ground-truth class count, a label-informed selection that leaks supervised information into an otherwise unsupervised generative pipeline.
  • block size for block diversity Db = not specified
    The block size in Equation (17) is never defined; the metric's values depend on this unspecified choice.
assumptions (3)
  • domain assumption Linear mixing model: Y = AX + epsilon
    Equation (1) assumes each pixel is a linear combination of endmembers with additive noise; the entire generation pipeline depends on this.
  • domain assumption VCA correctly extracts the true endmembers
    Section III-B uses VCA endmembers to initialize the decoder; if VCA fails, the abundance space is not physically meaningful.
  • ad hoc to paper The ground-truth class count equals the intrinsic number of endmembers
    Section IV-A sets d to the number of categories, a choice that has no physical justification and leaks labels.
invented entities (2)
  • point fidelity Fp
    purpose: Evaluation metric measuring cosine similarity of generated pixels to the most similar real pixel
    The metric is proposed in the paper and is not validated against human perception or downstream tasks; it is also structurally aligned with the model's linear decoder.
  • block diversity Db
    purpose: Evaluation metric measuring similarity of generated blocks to real blocks
    Proposed without external validation, and the block size is left unspecified.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Hyperspectral Image Generation with Unmixing Guided Diffusion Model." pith.science (2026). https://pith.science/paper/4YLV7GUV

@misc{pith2026250602601,
  author       = {Pith},
  title        = {Pith review of: Hyperspectral Image Generation with Unmixing Guided Diffusion Model},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4YLV7GUV}},
  note         = {Machine review of arXiv:2506.02601}
}
read the original abstract

We address hyperspectral image (HSI) synthesis, a problem that has garnered growing interest yet remains constrained by the conditional generative paradigms that limit sample diversity. While diffusion models have emerged as a state-of-the-art solution for high-fidelity image generation, their direct extension from RGB to hyperspectral domains is challenged by the high spectral dimensionality and strict physical constraints inherent to HSIs. To overcome the challenges, we introduce a diffusion framework explicitly guided by hyperspectral unmixing. The approach integrates two collaborative components: (i) an unmixing autoencoder that projects generation from the image domain into a low-dimensional abundance manifold, thereby reducing computational burden while maintaining spectral fidelity; and (ii) an abundance diffusion process that enforces non-negativity and sum-to-one constraints, ensuring physical consistency of the synthesized data. We further propose two evaluation metrics tailored to hyperspectral characteristics. Comprehensive experiments, assessed with both conventional measures and the proposed metrics, demonstrate that our method produces HSIs with both high quality and diversity, advancing the state of the art in hyperspectral data generation.

Figures

Figures reproduced from arXiv: 2506.02601 by the authors.

Figure 1
Figure 1. Components of HUD. for the diffusion module. To train the diffusion model, we first encode the hyperspectral images into the abundance space using the unmixing encoder, followed by a projection to relax the constraints of non-negativity and unity. The diffusion model is then trained to learn the distribution of the projected abundance maps. During sampling, we first sample from a standard Gaussian distribution, then… view at source ↗
Figure 2
Figure 2. Point fidelity and block diversity However, they require a classification model pretrained on large-scale datasets, which is not feasible for HSI due to the limited data size. Moreover, IS and FID are not designed to capture the unique characteristics of HSI data, such as the high spectral dimensionality and limited spatial resolution. Consequently, we propose two evaluation metrics tailored for HSI: point fidelity … view at source ↗
Figure 3
Figure 3. Pseudo-color visualization of real HSI and generated HSI. [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Spectral curve distributions of real HSI and generated HSI. [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

47 extracted references · 44 canonical work pages

  1. [1]

    Spectral discrepancy and cross-modal semantic consistency learning for object detection in hyperspectral images,

    X. He, C. Tang, X. Liu, W. Zhang, Z. Gao, C. Li, S. Qiu, and J. Xu, “Spectral discrepancy and cross-modal semantic consistency learning for object detection in hyperspectral images,” IEEE Transactions on Multimedia, pp. 1–13, 2025

    work page 2025

  2. [2]

    Gitanet: Group interactive threshold-based attention network for hyperspectral image classification,

    Y . Duan, C. Chen, M. Fu, X. Gong, Y . Niu, and F. Luo, “Gitanet: Group interactive threshold-based attention network for hyperspectral image classification,” IEEE Transactions on Multimedia, vol. 27, pp. 3571–3584, 2025

    work page 2025

  3. [3]

    Dual-stage approach toward hyperspectral image super-resolution,

    Q. Li, Y . Yuan, X. Jia, and Q. Wang, “Dual-stage approach toward hyperspectral image super-resolution,” IEEE Transactions on Image Processing, vol. 31, pp. 7252–7263, 2022

    work page 2022

  4. [4]

    From appearance to inherence: A hyperspectral image dataset and benchmark of material classification for surveillance,

    L. Gao, H.-M. Hu, X. Xue, and H. Hu, “From appearance to inherence: A hyperspectral image dataset and benchmark of material classification for surveillance,” IEEE Transactions on Multimedia , vol. 26, pp. 8569–8580, 2024

    work page 2024

  5. [5]

    Spectral Super-Resolution Meets Deep Learning: Achievements and Challenges,

    J. He, Q. Yuan, J. Li, Y . Xiao, D. Liu, H. Shen, and L. Zhang, “Spectral Super-Resolution Meets Deep Learning: Achievements and Challenges,” Information Fusion, vol. 97, p. 101812, 2023

    work page 2023

  6. [6]

    A Survey on Computational Spectral Reconstruction Methods from RGB to Hyperspectral Imaging,

    J. Zhang, R. Su, Q. Fu, W. Ren, F. Heide, and Y . Nie, “A Survey on Computational Spectral Reconstruction Methods from RGB to Hyperspectral Imaging,” Scientific Reports , vol. 12, no. 1, p. 11905, 2022

    work page 2022

  7. [7]

    Hierarchical context measurement net- work for single hyperspectral image super-resolution,

    H. Wang, C. Wang, and Y . Yuan, “Hierarchical context measurement net- work for single hyperspectral image super-resolution,” IEEE Transactions on Multimedia, vol. 27, pp. 2623–2637, 2025

    work page 2025

  8. [8]

    Knlconv: Kernel-space non-local convolution for hyperspectral image super-resolution,

    R. Ran, L.-J. Deng, T.-J. Zhang, J. Chang, X. Wu, and Q. Tian, “Knlconv: Kernel-space non-local convolution for hyperspectral image super-resolution,” IEEE Transactions on Multimedia , vol. 26, pp. 8836– 8848, 2024

    work page 2024

Show all 47 references
  1. [9]

    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,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2023, pp. 22 511–22 521

    2023

  2. [10]

    High-Quality and Diverse Few-Shot Image Generation via Masked Discrimination,

    J. Zhu, H. Ma, J. Chen, and J. Yuan, “High-Quality and Diverse Few-Shot Image Generation via Masked Discrimination,” IEEE Transactions on Image Processing, vol. 33, pp. 2950–2965, 2024

    2024

  3. [11]

    Auto-Encoding Variational Bayes,

    D. P. Kingma, “Auto-Encoding Variational Bayes,” arXiv preprint arXiv:1312.6114, 2013

    2013 arXiv

  4. [12]

    Taming Transformers for High- Resolution Image Synthesis,

    P. Esser, R. Rombach, and B. Ommer, “Taming Transformers for High- Resolution Image Synthesis,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2021, pp. 12 873–12 883. UNDER REVIEW 9

    2021

  5. [13]

    A glrt-based multi-pixel target detector in hyperspectral imagery,

    L. Chen, J. Liu, W. Chen, and B. Du, “A glrt-based multi-pixel target detector in hyperspectral imagery,” IEEE Transactions on Multimedia , vol. 25, pp. 2710–2722, 2023

    2023

  6. [14]

    Hyperspectral Image Denoising via Robust Subspace Estimation and Group Sparsity Constraint,

    X. Fu, Y . Guo, M. Xu, and S. Jia, “Hyperspectral Image Denoising via Robust Subspace Estimation and Group Sparsity Constraint,” IEEE Transactions on Geoscience and Remote Sensing , vol. 61, pp. 1–16, 2023

    2023

  7. [15]

    Hy- perspectral Umixing Based on Nonnegative Matrix Factorization: A Comprehensive Review,

    X.-R. Feng, H.-C. Li, R. Wang, Q. Du, X. Jia, and A. Plaza, “Hy- perspectral Umixing Based on Nonnegative Matrix Factorization: A Comprehensive Review,” IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing , vol. 15, pp. 4414–4436, 2022

    2022

  8. [16]

    Denoising Diffusion Probabilistic Models,

    J. Ho, A. Jain, and P. Abbeel, “Denoising Diffusion Probabilistic Models,” Advances in neural information processing systems , vol. 33, pp. 6840– 6851, 2020

    2020

  9. [17]

    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 2015: 18th International Conference, Munich, Germany, October 5-9, 2015, Proceedings, Part III

    2015

  10. [18]

    Springer, 2015, pp. 234–241

    2015

  11. [19]

    High-Resolution Image Synthesis with Latent Diffusion Models,

    R. Rombach, A. Blattmann, D. Lorenz, P. Esser, and B. Ommer, “High-Resolution Image Synthesis with Latent Diffusion Models,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 10 684–10 695

    2022

  12. [20]

    Hyperspectral Remote Sensing Imagery Generation from RGB Images Based on Joint Discrim- ination,

    L. Liu, S. Lei, Z. Shi, N. Zhang, and X. Zhu, “Hyperspectral Remote Sensing Imagery Generation from RGB Images Based on Joint Discrim- ination,” IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing , vol. 14, pp. 7624–7636, 2021

    2021

  13. [21]

    Whanet:wavelet- based hybrid asymmetric network for spectral super-resolution from rgb inputs,

    N. Wang, S. Mei, Y . Wang, Y . Zhang, and D. Zhan, “Whanet:wavelet- based hybrid asymmetric network for spectral super-resolution from rgb inputs,” IEEE Transactions on Multimedia , vol. 27, pp. 414–428, 2025

    2025

  14. [22]

    A Comprehensive Review on Deep Learning Based Remote Sensing Image Super-Resolution Methods,

    P. Wang, B. Bayram, and E. Sertel, “A Comprehensive Review on Deep Learning Based Remote Sensing Image Super-Resolution Methods,” Earth-Science Reviews, vol. 232, p. 104110, 2022

    2022

  15. [23]

    R2H-CCD: Hyperspectral Imagery Generation from RGB Images Based on Conditional Cascade Diffusion Probabilistic Models,

    L. Zhang, X. Luo, S. Li, and X. Shi, “R2H-CCD: Hyperspectral Imagery Generation from RGB Images Based on Conditional Cascade Diffusion Probabilistic Models,” in IGARSS 2023-2023 IEEE International Geo- science and Remote Sensing Symposium . IEEE, 2023, pp. 7392–7395

    2023

  16. [24]

    Spectral super-resolution network guided by intrinsic properties of hyperspectral imagery,

    R. Hang, Q. Liu, and Z. Li, “Spectral super-resolution network guided by intrinsic properties of hyperspectral imagery,” IEEE Transactions on Image Processing, vol. 30, pp. 7256–7265, 2021

    2021

  17. [25]

    DCN-T: Dual Context Network with Transformer for Hyperspectral Image Classification,

    D. Wang, J. Zhang, B. Du, L. Zhang, and D. Tao, “DCN-T: Dual Context Network with Transformer for Hyperspectral Image Classification,” IEEE Transactions on Image Processing , vol. 32, pp. 2536–2551, 2023

    2023

  18. [26]

    SpectralDiff: A Generative Framework for Hyperspectral Image Classification with Diffusion Models,

    N. Chen, J. Yue, L. Fang, and S. Xia, “SpectralDiff: A Generative Framework for Hyperspectral Image Classification with Diffusion Models,” IEEE Transactions on Geoscience and Remote Sensing , 2023

    2023

  19. [27]

    Generative Adversarial Network with Transformer for Hyperspectral Image Classification,

    S. Hao, Y . Xia, and Y . Ye, “Generative Adversarial Network with Transformer for Hyperspectral Image Classification,” IEEE Geoscience and Remote Sensing Letters , 2023

    2023

  20. [28]

    HSIGENE: A Foundation Model for Hyperspectral Image Generation,

    L. Pang, X. Cao, D. Tang, S. Xu, X. Bai, F. Zhou, and D. Meng, “HSIGENE: A Foundation Model for Hyperspectral Image Generation,” arXiv preprint arXiv:2409.12470 , 2024

    2024 arXiv

  21. [29]

    Diverse Hyperspec- tral Remote Sensing Image Synthesis with Diffusion Models,

    L. Liu, B. Chen, H. Chen, Z. Zou, and Z. Shi, “Diverse Hyperspec- tral Remote Sensing Image Synthesis with Diffusion Models,” IEEE Transactions on Geoscience and Remote Sensing , vol. 61, pp. 1–16, 2023

    2023

  22. [30]

    UnmixDiff: Umixing-Based Diffusion Model for Hyperspectral Image Synthesis,

    Y . Yu, E. Pan, Y . Ma, X. Mei, Q. Chen, and J. Ma, “UnmixDiff: Umixing-Based Diffusion Model for Hyperspectral Image Synthesis,” IEEE Transactions on Geoscience and Remote Sensing , vol. 62, pp. 1–18, 2024

    2024

  23. [31]

    HSIGAN: A Conditional Hyperspectral Image Synthesis Method with Auxiliary Classifier,

    W. Liu, J. You, and J. Lee, “HSIGAN: A Conditional Hyperspectral Image Synthesis Method with Auxiliary Classifier,” IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing , vol. 14, pp. 3330–3344, 2021

    2021

  24. [32]

    Physics-informed Hyperspectral Remote Sensing Image Synthesis With Deep Conditional Generative Adversarial Networks,

    L. Liu, W. Li, Z. Shi, and Z. Zou, “Physics-informed Hyperspectral Remote Sensing Image Synthesis With Deep Conditional Generative Adversarial Networks,” IEEE Transactions on Geoscience and Remote Sensing, vol. 60, pp. 1–15, 2022

    2022

  25. [33]

    Specdm: Hyperspectral Dataset Synthesis with Pixel-level Semantic Annotations,

    W. Liu, P. Yang, W. Hong, X. Mei, and J. Ma, “Specdm: Hyperspectral Dataset Synthesis with Pixel-level Semantic Annotations,” 2025. [Online]. Available: https://arxiv.org/abs/2502.17056

    2025 arXiv

  26. [34]

    From the Abundance Perspective: Multi-modal Scene Fusion-based Hyperspectral Image Synthesis,

    E. Pan, Y . Yu, X. Mei, J. Huang, and J. Ma, “From the Abundance Perspective: Multi-modal Scene Fusion-based Hyperspectral Image Synthesis,” Information Fusion , vol. 108, p. 102419, 2024. [Online]. Available: https://www.sciencedirect.com/science/article/pii/ S1566253524001970

    2024

  27. [35]

    Hyperspectral Umixing Overview: Geometrical, Statistical, and Sparse Regression-Based Approaches,

    J. M. Bioucas-Dias, A. Plaza, N. Dobigeon, M. Parente, Q. Du, P. Gader, and J. Chanussot, “Hyperspectral Umixing Overview: Geometrical, Statistical, and Sparse Regression-Based Approaches,” IEEE journal of selected topics in applied earth observations and remote sensing , vol....

    2012

  28. [36]

    Spectral Variability in Hyperspectral Data Umixing: A Comprehensive Review,

    R. A. Borsoi, T. Imbiriba, J. C. M. Bermudez, C. Richard, J. Chanussot, L. Drumetz, J.-Y . Tourneret, A. Zare, and C. Jutten, “Spectral Variability in Hyperspectral Data Umixing: A Comprehensive Review,” IEEE geoscience and remote sensing magazine , vol. 9, no. 4, pp. 223–270, 2021

    2021

  29. [37]

    Vertex Component Analysis: A Fast Algorithm to Unmix Hyperspectral Data,

    J. M. Nascimento and J. M. Dias, “Vertex Component Analysis: A Fast Algorithm to Unmix Hyperspectral Data,” IEEE Transactions on Geoscience and Remote Sensing , vol. 43, no. 4, pp. 898–910, 2005

    2005

  30. [38]

    Spectral-Spatial Hyperspectral Umixing Using Nonnegative Matrix Factorization,

    S. Zhang, G. Zhang, F. Li, C. Deng, S. Wang, A. Plaza, and J. Li, “Spectral-Spatial Hyperspectral Umixing Using Nonnegative Matrix Factorization,” IEEE Transactions on Geoscience and Remote Sensing , vol. 60, pp. 1–13, 2022

    2022

  31. [39]

    Partial NMF-Based Hyperspectral Unmixing Methods for Linear Mixing Models Addressing Intra-Class Variability,

    M. Iftene, F. Z. Benhalouche, Y . K. Benkouider, and M. S. Karoui, “Partial NMF-Based Hyperspectral Unmixing Methods for Linear Mixing Models Addressing Intra-Class Variability,” Digital Signal Processing , vol. 143, p. 104264, 2023

    2023

  32. [40]

    Efficient hyperspectral sparse regression unmixing with multilayers,

    X. Shen, L. Chen, H. Liu, X. Su, W. Wei, X. Zhu, and X. Zhou, “Efficient hyperspectral sparse regression unmixing with multilayers,” IEEE Transactions on Geoscience and Remote Sensing , vol. 61, pp. 1–14, 2023

    2023

  33. [41]

    Suncnn: Sparse unmixing using unsupervised convolutional neural network,

    B. Rasti and B. Koirala, “Suncnn: Sparse unmixing using unsupervised convolutional neural network,” IEEE Geoscience and Remote Sensing Letters, vol. 19, pp. 1–5, 2021

    2021

  34. [42]

    Deep Autoencoder for Hy- perspectral Unmixing via Global-Local Smoothing,

    X. Xu, X. Song, T. Li, Z. Shi, and B. Pan, “Deep Autoencoder for Hy- perspectral Unmixing via Global-Local Smoothing,” IEEE Transactions on Geoscience and Remote Sensing , vol. 60, pp. 1–16, 2022

    2022

  35. [43]

    A Reversible Generative Network for Hyperspectral Umixing with Spectral Variability,

    Y . Gao, B. Pan, X. Xu, X. Song, and Z. Shi, “A Reversible Generative Network for Hyperspectral Umixing with Spectral Variability,” IEEE Transactions on Geoscience and Remote Sensing , 2024

    2024

  36. [44]

    Endmember-Guided Umixing Network (EGU-Net): A General Deep Learning Framework for Self-Supervised Hyperspectral Umixing,

    D. Hong, L. Gao, J. Yao, N. Yokoya, J. Chanussot, U. Heiden, and B. Zhang, “Endmember-Guided Umixing Network (EGU-Net): A General Deep Learning Framework for Self-Supervised Hyperspectral Umixing,” IEEE Transactions on Neural Networks and Learning Systems , vol. 33, no. 11, pp...

    2022

  37. [45]

    Improved Training of Wasserstein GANs,

    I. Gulrajani, F. Ahmed, M. Arjovsky, V . Dumoulin, and A. C. Courville, “Improved Training of Wasserstein GANs,” Advances in neural informa- tion processing systems , vol. 30, 2017

    2017

  38. [46]

    Multi-Stage Progressive Image Restoration,

    S. W. Zamir, A. Arora, S. Khan, M. Hayat, F. S. Khan, M.-H. Yang, and L. Shao, “Multi-Stage Progressive Image Restoration,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), June 2021, pp. 14 821–14 831

    2021

  39. [47]

    Unmixing Before Fusion: A Generalized Paradigm for Multi-Source-Based Hyperspectral Image Synthesis,

    Y . Yu, E. Pan, X. Wang, Y . Wu, X. Mei, and J. Ma, “Unmixing Before Fusion: A Generalized Paradigm for Multi-Source-Based Hyperspectral Image Synthesis,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , June 2024, pp. 9297– 9306. S...

    2024

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.