pith. sign in

arxiv: 2606.03806 · v1 · pith:EOU36ZDTnew · submitted 2026-06-02 · 💻 cs.CV

TeX-1500: A Paired Real-World LWIR Hyperspectral Dataset and Benchmark for Temperature-Emissivity-Texture Decomposition

Cheng Dai , Jiale Lin , Hongyi Xu , Bingxuan Song , Ziyang Xie , Fanglin Bao This is my paper

Pith reviewed 2026-06-28 10:35 UTC · model grok-4.3

classification 💻 cs.CV
keywords LWIR hyperspectral imagingtemperature-emissivity-texture decompositionpaired datasetbenchmarkthermal perceptionhyperspectral decompositionreal-world datasetsupervised learning
0
0 comments X

The pith

TeX-1500 supplies 1522 real paired LWIR hyperspectral scenes with aligned temperature, emissivity, and texture labels.

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

The paper presents TeX-1500 as a large collection of calibrated real-world LWIR hyperspectral image pairs that include corresponding temperature, emissivity, and texture fields. Existing decomposition methods have relied on scene-specific solvers because no large paired dataset existed for training data-driven models. The new dataset draws from multiple locations, seasons, times, and two sensor families, with labels built through a fixed restoration and construction protocol. A baseline wavelength-aware network is included to demonstrate that models can learn the mapping from radiance cubes to the decomposed fields. The result is a benchmark that supports supervised training and evaluation for recovering physical properties from thermal hyperspectral data.

Core claim

TeX-1500 contains 1,522 calibrated real-scene pairs of LWIR hyperspectral imagery and their temperature-emissivity-texture decompositions, drawn from DARPA IH pushbroom and FTIR acquisitions across five locations and four seasons. Each pair stores a valid-band radiance cube, wavelength positions, and aligned supervision labels produced by a consistent restoration and TeX-construction protocol. A simple TeX-UNet baseline maps the input bands and wavelengths to the three output fields, and experiments on held-out scenes plus zero- and few-shot transfer confirm that the pairs supply usable supervision for data-driven decomposition.

What carries the argument

The TeX-1500 paired dataset, which supplies real LWIR hyperspectral radiance cubes together with constructed temperature, emissivity, and texture supervision.

If this is right

  • Supervised models can be trained end-to-end to map LWIR hyperspectral cubes directly to temperature, emissivity, and texture fields.
  • Data-driven decomposition becomes feasible as an alternative to per-scene inverse solvers.
  • Performance of different decomposition algorithms can be compared quantitatively on the same held-out real scenes.
  • Models trained on the dataset can be tested for transfer to new sensors and wavelength layouts without retraining from scratch.
  • Physical-property recovery in thermal perception can shift from purely optimization-based methods to learned ones.

Where Pith is reading between the lines

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

  • The dataset construction protocol could be reused or adapted to create similar paired supervision in other spectral ranges or imaging modalities.
  • Models trained on TeX-1500 might improve downstream tasks such as material classification or anomaly detection that rely on accurate emissivity or temperature estimates.
  • The benchmark could encourage development of architectures that explicitly handle variable numbers of spectral bands.
  • Larger-scale collection using the same protocol might reduce reliance on any single sensor family.

Load-bearing premise

The temperature, emissivity, and texture labels built by the consistent restoration and TeX-construction protocol are accurate enough to act as reliable supervision.

What would settle it

Independent high-precision measurements on a subset of the same scenes that show systematic, repeatable differences from the dataset's constructed TeX labels.

Figures

Figures reproduced from arXiv: 2606.03806 by Bingxuan Song, Cheng Dai, Fanglin Bao, Hongyi Xu, Jiale Lin, Ziyang Xie.

Figure 1
Figure 1. Figure 1: Acquisition-time distribution of TeX-1500. Samples span diverse daytime, nighttime, and transitional thermal conditions, capturing variations in object heat state, thermal contrast, and downwelling environmental radiance. downwelling radiative transfer to restore degraded LWIR HSI. These solvers establish the physical basis of TeX decomposi￾tion, but their per-scene optimization depends on choices such as … view at source ↗
Figure 2
Figure 2. Figure 2: Spectral coverage of TeX-1500. DARPA IH pushbroom and FTIR observations occupy thermal-infrared wavelength ranges with overlapping LWIR atmospheric-window coverage, while retaining sensor-specific band limits, sampling densities, and valid-band layouts [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Semantic-class distribution of TeX-1500. The DARPA IH subset (left) and FTIR subset (right) are summarized by their top-7 category proportions. samples through a common construction pipeline. The DARPA IH train/validation/test split uses held-out scenes and acquisition dates to test geographic and temporal generalization, while the FTIR split emphasizes changes in sensor layout, scene content, and material… view at source ↗
Figure 5
Figure 5. Figure 5: TeX-UNet baseline for HSI-to-TeX inversion. The model takes calibrated HSI bands and their wavelength positions as input and predicts temperature T, emissivity e, and texture X. selection, and report the DARPA IH test split in Table III. During training, each sample is formed by a random 224×224 spatial crop and a random selection of 64 valid spectral bands with their calibrated wavelength positions. Spati… view at source ↗
Figure 4
Figure 4. Figure 4: TeX label (right) visual quality compared with HADAR-SGD (left). D. Dataset learnability assessment We further evaluate TeX-1500 as paired supervision for learning-based HSI-to-TeX inversion. As an initial baseline, we train TeX-UNet4 ( [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figure 6
Figure 6. Figure 6: TeX-UNet results on the DARPA IH test split. The model is trained on the DARPA IH training split and evaluated on DARPA IH test scenes. The predicted TeX maps (left) preserve the main temperature, emissivity, and texture structures of the ground-truth labels (right). TABLE II TEX-UNET TRAINING HYPERPARAMETERS. Setting Value Training input Random 224 × 224 crop and sampling of 64 valid bands Spectral encodi… view at source ↗
Figure 8
Figure 8. Figure 8: Training set samples from TeX-1500 DARPA IH pushbroom subset at Sidewinder Range, TPG, AZ in August 2021. Panels show T (in K), e, X, and HSI-band radiance (W · m−2 · sr−1 · µm−1 ) [PITH_FULL_IMAGE:figures/full_fig_p008_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Training set samples from TeX-1500 DARPA IH pushbroom subset at Loring Commerce Center, ME in December 2021. Panels show T (in K), e, X, and HSI-band radiance (W · m−2 · sr−1 · µm−1 ) [PITH_FULL_IMAGE:figures/full_fig_p009_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Training set samples from TeX-1500 DARPA IH pushbroom subset at Avon Park Air Force Range, FL in April 2022. Panels show T (in K), e, X, and HSI-band radiance (W · m−2 · sr−1 · µm−1 ) [PITH_FULL_IMAGE:figures/full_fig_p010_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Validation set samples from TeX-1500 DARPA IH pushbroom subsetat Sidewinder Range, TPG, AZ in September 2020. Panels show T (in K), e, X, and HSI-band radiance (W · m−2 · sr−1 · µm−1 ) [PITH_FULL_IMAGE:figures/full_fig_p011_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Testing set samples from TeX-1500 DARPA IH pushbroom subset at Fort A. P. Hill, VA in April 2021. Panels show T (in K), e, X, and HSI-band radiance (W · m−2 · sr−1 · µm−1 ) [PITH_FULL_IMAGE:figures/full_fig_p012_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: Training set samples from TeX-1500 FTIR subset. Panels show T (in K), e, X, and HSI-band radiance (W · m−2 · sr−1 · µm−1 ) [PITH_FULL_IMAGE:figures/full_fig_p013_13.png] view at source ↗
read the original abstract

Temperature-emissivity-texture (TeX) decomposition seeks to recover object heat state, material spectral response, and visible-like geometric texture from long-wave infrared hyperspectral imaging (LWIR HSI). Existing TeX pipelines are mainly scene-specific inverse solvers, and the lack of paired LWIR HSI-TeX supervision has limited learning-based decomposition. To address this gap, we introduce TeX-1500, a large-scale paired LWIR HSI-TeX dataset and benchmark for supervised HSI-to-TeX decomposition. TeX-1500 contains 1,522 calibrated real-scene pairs from DARPA Invisible Headlights (DARPA IH) pushbroom imagery and our FTIR acquisitions, covering five locations, four seasons, diverse acquisition times, heterogeneous wavelength layouts, and two sensor families. Each sample stores a calibrated valid-band radiance cube, calibrated wavelength positions, and aligned temperature, emissivity, and texture supervision constructed through a consistent restoration and TeX-construction protocol. We further provide TeX-UNet, a simple wavelength-aware baseline that maps calibrated HSI bands and wavelength positions to TeX fields. Experiments on the held-out DARPA IH pushbroom scenes and zero-/few-shot transfer to FTIR scenes show that TeX-1500 provides usable paired supervision and a measurable benchmark for data-driven physical-property-centered thermal perception.

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

1 major / 2 minor

Summary. The paper introduces TeX-1500, a dataset of 1,522 real-world paired LWIR hyperspectral images and aligned temperature-emissivity-texture (TeX) fields constructed from DARPA IH pushbroom and FTIR data across diverse scenes, seasons, and sensors. It describes a consistent restoration and TeX-construction protocol for generating the supervision, releases a wavelength-aware TeX-UNet baseline, and reports held-out and transfer experiments to argue that the dataset enables supervised learning for physical-property-centered thermal perception.

Significance. If the constructed TeX labels are shown to be accurate, the contribution would be significant: it supplies the first large-scale, real-scene paired supervision for data-driven TeX decomposition, where prior work relied on scene-specific inverse solvers. The diversity of acquisition conditions and the inclusion of zero-/few-shot transfer results to a second sensor family strengthen the benchmark value. The release of calibrated radiance cubes, wavelength metadata, and a reproducible baseline further supports downstream use.

major comments (1)
  1. [Dataset construction / TeX-construction protocol] Section describing the TeX-construction protocol (referenced in the abstract and dataset section): the central claim that TeX-1500 supplies 'usable paired supervision' rests on the accuracy of the constructed temperature, emissivity, and texture fields, yet the manuscript provides no quantitative error characterization, cross-validation against independent physical measurements (e.g., contact thermometry or calibrated blackbody references), or sensitivity analysis of residual calibration/inversion errors. This is load-bearing because unquantified systematic bias in the labels would render the benchmark self-referential rather than an external test of learned decomposition.
minor comments (2)
  1. [Dataset description] The description of wavelength layouts and valid-band selection across the two sensor families could be expanded with an explicit table or figure showing band counts and spectral coverage per subset.
  2. [Figures] Figure captions for the example TeX fields should include the specific scene ID, acquisition time, and sensor to allow direct correspondence with the reported transfer results.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive review and for emphasizing the need to substantiate the accuracy of the constructed TeX labels. We address the major comment below and will revise the manuscript accordingly.

read point-by-point responses

  1. Referee: [Dataset construction / TeX-construction protocol] Section describing the TeX-construction protocol (referenced in the abstract and dataset section): the central claim that TeX-1500 supplies 'usable paired supervision' rests on the accuracy of the constructed temperature, emissivity, and texture fields, yet the manuscript provides no quantitative error characterization, cross-validation against independent physical measurements (e.g., contact thermometry or calibrated blackbody references), or sensitivity analysis of residual calibration/inversion errors. This is load-bearing because unquantified systematic bias in the labels would render the benchmark self-referential rather than an external test of learned decomposition.

    Authors: We agree that quantitative characterization of label accuracy is essential to support the claim of usable paired supervision. The current manuscript details the restoration and TeX-construction protocol but does not include error metrics or sensitivity analysis. In the revised version we will add a dedicated subsection that (i) propagates reported calibration uncertainties from the DARPA IH pushbroom and FTIR source data into the derived temperature, emissivity, and texture fields, (ii) presents a sensitivity study over key inversion parameters (atmospheric correction, emissivity priors, and band selection), and (iii) compares the constructed fields against the blackbody reference measurements that were part of the original DARPA IH and FTIR calibration protocols. While independent contact thermometry was not collected during the field campaigns, the added analysis will quantify residual errors and bound potential systematic bias, thereby addressing the concern that the benchmark could be self-referential. revision: yes

Circularity Check

0 steps flagged

No circularity: dataset construction with external sources and no self-referential derivation

full rationale

The paper introduces TeX-1500 as a paired dataset constructed from DARPA IH and FTIR acquisitions via a described restoration protocol, with a simple baseline model TeX-UNet. No mathematical derivation chain, fitted parameters renamed as predictions, or self-citation load-bearing uniqueness theorems exist. The contribution is the dataset and benchmark itself rather than any prediction that reduces to its inputs by construction. The protocol is applied to independent external data, and no equations or self-referential steps are present that would trigger the enumerated circularity patterns.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the assumption that the TeX labels are faithful; no free parameters or invented physical entities are introduced.

axioms (1)
  • domain assumption The restoration and TeX-construction protocol produces accurate temperature, emissivity, and texture supervision from the radiance cubes.
    This premise is required for the paired supervision to be usable; it is invoked when the abstract states the labels are 'constructed through a consistent restoration and TeX-construction protocol'.

pith-pipeline@v0.9.1-grok · 5796 in / 1231 out tokens · 26658 ms · 2026-06-28T10:35:08.190870+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

49 extracted references · 7 canonical work pages · 1 internal anchor

  1. [1]

    Learning Continuous Wasserstein Barycenter Space for Generalized All-in-One Image Restoration ,

    X. Tang, X. He, J. Xu, X. Gu, and J. Sun, “ Learning Continuous Wasserstein Barycenter Space for Generalized All-in-One Image Restoration ,”IEEE Transactions on Pattern Analysis & Machine Intelligence, no. 01, pp. 1–16, Feb. 5555. [Online]. Available: https://doi.ieeecomputersociety.org/10.1109/TPAMI.2026.3669121

    work page doi:10.1109/tpami.2026.3669121 2026

  2. [2]

    Single image haze removal using dark channel prior,

    K. He, J. Sun, and X. Tang, “Single image haze removal using dark channel prior,”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 33, no. 12, pp. 2341–2353, 2011

    2011

  3. [3]

    Visual-in-Visual: A Unified and Efficient Baseline for Image Restoration ,

    Y . Cui, W. Ren, B. Shi, and A. Knoll, “ Visual-in-Visual: A Unified and Efficient Baseline for Image Restoration ,”IEEE Transactions on Pattern Analysis & Machine Intelligence, no. 01, pp. 1–18, Mar

  4. [4]

    Available: https://doi.ieeecomputersociety.org/10.1109/ TPAMI.2026.3669720

    [Online]. Available: https://doi.ieeecomputersociety.org/10.1109/ TPAMI.2026.3669720

    work page arXiv 2026

  5. [5]

    Heat-assisted detection and ranging,

    F. Bao, X. Wang, S. H. Sureshbabu, G. Sreekumar, L. Yang, V . Aggar- wal, V . N. Boddeti, and Z. Jacob, “Heat-assisted detection and ranging,” Nature, vol. 619, no. 7971, pp. 743–748, 2023

    2023

  6. [6]

    Seeing through fog without seeing fog: Deep multi- modal sensor fusion in unseen adverse weather,

    M. Bijelic, T. Gruber, F. Mannan, F. Kraus, W. Ritter, K. Dietmayer, and F. Heide, “Seeing through fog without seeing fog: Deep multi- modal sensor fusion in unseen adverse weather,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020, pp. 11 682–11 692

    2020

  7. [7]

    Rethinking data augmentation for robust lidar semantic segmentation in adverse weather,

    J. Park, K. Kim, and H. Shim, “Rethinking data augmentation for robust lidar semantic segmentation in adverse weather,” inComputer Vision – ECCV 2024. Springer, 2024, pp. 320–336

    2024

  8. [8]

    Deep learning- based robust positioning for all-weather autonomous driving,

    Y . Almalioglu, M. Turan, N. Trigoni, and A. Markham, “Deep learning- based robust positioning for all-weather autonomous driving,”Nature machine intelligence, vol. 4, no. 9, pp. 749–760, 2022

    2022

  9. [9]

    Non-line-of-sight imaging with picosecond temporal resolution,

    B. Wang, M.-Y . Zheng, J.-J. Han, X. Huang, X.-P. Xie, F. Xu, Q. Zhang, and J.-W. Pan, “Non-line-of-sight imaging with picosecond temporal resolution,”Physical Review Letters, vol. 127, no. 5, p. 053602, 2021

    2021

  10. [10]

    Real-time non-line-of-sight computational imaging using spectrum filtering and motion compensation,

    J.-T. Ye, Y . Sun, W. Li, J.-W. Zeng, Y . Hong, Z.-P. Li, X. Huang, X. Xue, X. Yuan, F. Xuet al., “Real-time non-line-of-sight computational imaging using spectrum filtering and motion compensation,”Nature Computational Science, vol. 4, no. 12, pp. 920–927, 2024

    2024

  11. [11]

    Why thermal images are blurry,

    F. Bao, S. Jape, A. Schramka, J. Wang, T. E. McGraw, and Z. Jacob, “Why thermal images are blurry,”Optics Express, vol. 32, no. 3, pp. 3852–3865, 2024

    2024

  12. [12]

    HADAR-Based Thermal Infrared Hyperspectral Image Restoration

    C. Dai, J. Lin, B. Song, Y . Chen, J. Chen, X. Yuan, and F. Bao, “Hadar-based thermal infrared hyperspectral image restoration,” 2026. [Online]. Available: https://arxiv.org/abs/2605.13664

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  13. [13]

    Universal computational thermal imaging overcoming the ghosting effect,

    H. Xu, D. Wang, C. Zhao, J. Chen, J. Lin, L. Cao, Y . Zhong, Y . She, and F. Bao, “Universal computational thermal imaging overcoming the ghosting effect,” 2026. [Online]. Available: https: //arxiv.org/abs/2604.01542

    work page arXiv 2026

  14. [14]

    Concurrent band selection and traversability estimation from long-wave hyperspec- tral imagery in off-road settings,

    F. Yellin, S. McCloskey, C. Hill, E. Smith, and B. Clipp, “Concurrent band selection and traversability estimation from long-wave hyperspec- tral imagery in off-road settings,” inProceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, 2024, pp. 7483– 7492

    2024

  15. [15]

    Generalized assorted pixel camera: postcapture control of resolution, dynamic range, and spectrum,

    F. Yasuma, T. Mitsunaga, D. Iso, and S. K. Nayar, “Generalized assorted pixel camera: postcapture control of resolution, dynamic range, and spectrum,”IEEE transactions on image processing, vol. 19, no. 9, pp. 2241–2253, 2010

    2010

  16. [16]

    Statistics of real-world hyperspectral images,

    A. Chakrabarti and T. Zickler, “Statistics of real-world hyperspectral images,” inCVPR 2011, 2011, pp. 193–200

    2011

  17. [17]

    Sparse recovery of hyperspectral signal from natural rgb images,

    B. Arad and O. Ben-Shahar, “Sparse recovery of hyperspectral signal from natural rgb images,” inEuropean conference on computer vision. Springer, 2016, pp. 19–34

    2016

  18. [18]

    Ntire 2022 spectral recovery challenge and data set,

    B. Arad, R. Timofte, R. Yahel, N. Morag, A. Bernat, Y . Cai, J. Lin, Z. Lin, H. Wang, Y . Zhanget al., “Ntire 2022 spectral recovery challenge and data set,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 863–881

    2022

  19. [19]

    Hyperspectral and lidar data fusion: Outcome of the 2013 grss data fusion contest,

    C. Debes, A. Merentitis, R. Heremans, J. Hahn, N. Frangiadakis, T. van Kasteren, W. Liao, R. Bellens, A. Pi ˇzurica, S. Gautama, W. Philips, S. Prasad, Q. Du, and F. Pacifici, “Hyperspectral and lidar data fusion: Outcome of the 2013 grss data fusion contest,”IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, vol. 7, no. 6, ...

    2013

  20. [20]

    Whu-hi: Uav-borne hy- perspectral with high spatial resolution (h2) benchmark datasets for hyperspectral image classification,

    X. Hu, Y . Zhong, C. Luo, and X. Wang, “Whu-hi: Uav-borne hy- perspectral with high spatial resolution (h2) benchmark datasets for hyperspectral image classification,”arXiv preprint arXiv:2012.13920, 2020

    work page arXiv 2012

  21. [21]

    Satmae: Pre-training transformers for temporal and multi-spectral satellite imagery,

    Y . Cong, S. Khanna, C. Meng, P. Liu, E. Rozi, Y . He, M. Burke, D. Lo- bell, and S. Ermon, “Satmae: Pre-training transformers for temporal and multi-spectral satellite imagery,”Advances in Neural Information Processing Systems, vol. 35, pp. 197–211, 2022

    2022

  22. [22]

    Spectralgpt: Spectral remote sensing foun- dation model,

    D. Hong, B. Zhang, X. Li, Y . Li, C. Li, J. Yao, N. Yokoya, H. Li, P. Ghamisi, X. Jiaet al., “Spectralgpt: Spectral remote sensing foun- dation model,”IEEE transactions on pattern analysis and machine intelligence, vol. 46, no. 8, pp. 5227–5244, 2024

    2024

  23. [23]

    Spectralearth: Training hyperspectral foundation models at scale,

    N. A. A. Braham, C. M. Albrecht, J. Mairal, J. Chanussot, Y . Wang, and X. X. Zhu, “Spectralearth: Training hyperspectral foundation models at scale,”IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2025

    2025

  24. [24]

    Hypersigma: Hyperspectral intelligence comprehension foundation model,

    D. Wang, M. Hu, Y . Jin, Y . Miao, J. Yang, Y . Xu, X. Qin, J. Ma, L. Sun, C. Li, C. Fu, H. Chen, C. Han, N. Yokoya, J. Zhang, M. Xu, L. Liu, L. Zhang, C. Wu, B. Du, D. Tao, and L. Zhang, “Hypersigma: Hyperspectral intelligence comprehension foundation model,”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 47, no. 8, pp. 6427–6444, 2025

    2025

  25. [25]

    Multispectral pedestrian detection: Benchmark dataset and baseline,

    S. Hwang, J. Park, N. Kim, Y . Choi, and I. So Kweon, “Multispectral pedestrian detection: Benchmark dataset and baseline,” inProceedings of the IEEE conference on computer vision and pattern recognition, 2015, pp. 1037–1045

    2015

  26. [26]

    Pedestrian detection at day/night time with visible and fir cameras: A comparison,

    A. Gonz ´alez, Z. Fang, Y . Socarras, J. Serrat, D. V ´azquez, J. Xu, and A. M. L ´opez, “Pedestrian detection at day/night time with visible and fir cameras: A comparison,”Sensors, vol. 16, no. 6, p. 820, 2016

    2016

  27. [27]

    The tno multiband image data collection,

    A. Toet, “The tno multiband image data collection,”Data in brief, vol. 15, p. 249, 2017

    2017

  28. [28]

    Mfnet: Towards real-time semantic segmentation for autonomous vehicles with multi-spectral scenes,

    Q. Ha, K. Watanabe, T. Karasawa, Y . Ushiku, and T. Harada, “Mfnet: Towards real-time semantic segmentation for autonomous vehicles with multi-spectral scenes,” in2017 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2017, pp. 5108–5115

    2017

  29. [29]

    Heatnet: Bridging the day-night domain gap in semantic segmentation with thermal images,

    J. Vertens, J. Z ¨urn, and W. Burgard, “Heatnet: Bridging the day-night domain gap in semantic segmentation with thermal images,” in2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2020, pp. 8461–8468

    2020

  30. [30]

    Llvip: A visible-infrared paired dataset for low-light vision,

    X. Jia, C. Zhu, M. Li, W. Tang, and W. Zhou, “Llvip: A visible-infrared paired dataset for low-light vision,” inProceedings of the IEEE/CVF international conference on computer vision, 2021, pp. 3496–3504

    2021

  31. [31]

    Long-range uav thermal geo-localization with satellite imagery,

    J. Xiao, D. Tortei, E. Roura, and G. Loianno, “Long-range uav thermal geo-localization with satellite imagery,” in2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2023, pp. 5820–5827

    2023

  32. [32]

    Piafusion: A progres- sive infrared and visible image fusion network based on illumination aware,

    L. Tang, J. Yuan, H. Zhang, X. Jiang, and J. Ma, “Piafusion: A progres- sive infrared and visible image fusion network based on illumination aware,”Information Fusion, vol. 83, pp. 79–92, 2022

    2022

  33. [33]

    Mask-difuser: A masked diffusion model for unified unsupervised image fusion,

    L. Tang, C. Li, and J. Ma, “Mask-difuser: A masked diffusion model for unified unsupervised image fusion,”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 48, no. 1, pp. 591–608, 2026

    2026

  34. [34]

    Ddfm: Denoising diffusion model for multi-modality image fusion,

    Z. Zhao, H. Bai, Y . Zhu, J. Zhang, S. Xu, Y . Zhang, K. Zhang, D. Meng, R. Timofte, and L. Van Gool, “Ddfm: Denoising diffusion model for multi-modality image fusion,” inProceedings of the IEEE/CVF international conference on computer vision, 2023, pp. 8082–8093

    2023

  35. [35]

    Thermalgen: Style-disentangled flow-based generative models for rgb-to-thermal im- age translation,

    J. Xiao, R. Nayak, N. Zhang, D. Tortei, and G. Loianno, “Thermalgen: Style-disentangled flow-based generative models for rgb-to-thermal im- age translation,”Advances in Neural Information Processing Systems, vol. 38, pp. 33 111–33 133, 2026

    2026

  36. [36]

    Processing of multires- olution thermal hyperspectral and digital color data: Outcome of the 2014 ieee grss data fusion contest,

    W. Liao, X. Huang, F. Van Coillie, S. Gautama, A. Pi ˇzurica, W. Philips, H. Liu, T. Zhu, M. Shimoni, G. Moseret al., “Processing of multires- olution thermal hyperspectral and digital color data: Outcome of the 2014 ieee grss data fusion contest,”IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, vol. 8, no. 6, pp. 2984–2996, 2015

    2014

  37. [37]

    Spectral noise resistance split window atmospheric compensation for airborne thermal infrared hyperspectral,

    D. Wang, L. Cao, L. Gao, F. Ye, and Y . Zhong, “Spectral noise resistance split window atmospheric compensation for airborne thermal infrared hyperspectral,” inIGARSS 2025-2025 IEEE International Geoscience and Remote Sensing Symposium. IEEE, 2025, pp. 1244–1248

    2025

  38. [38]

    Toward noise-resilient retrieval of land surface temperature and emissivity using airborne thermal infrared hyperspectral imagery,

    D. Wang, L.-Q. Cao, Y .-H. Du, H.-Y . Xiong, F.-W. Ye, and Y .-F. Zhong, “Toward noise-resilient retrieval of land surface temperature and emissivity using airborne thermal infrared hyperspectral imagery,”ISPRS Journal of Photogrammetry and Remote Sensing, vol. 231, pp. 532–551, 2026

    2026

  39. [39]

    Absorption-based, passive range imaging from hyperspectral thermal measurements,

    U. Dorken Gallastegi, H. Rueda-Chac ´on, M. J. Stevens, and V . K. Goyal, “Absorption-based, passive range imaging from hyperspectral thermal measurements,”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 47, no. 5, pp. 4044–4060, 2025

    2025

  40. [40]

    Lwir hyperspectral image classification based on a temperature-emissivity residual network and conditional random field model,

    L. Cao, J. He, L. Gao, Y . Zhong, X. Hu, and Z. Li, “Lwir hyperspectral image classification based on a temperature-emissivity residual network and conditional random field model,”International Journal of Remote Sensing, vol. 43, no. 10, pp. 3744–3768, 2022. PREPRINT 2026 15

    2022

  41. [41]

    Physics-integrated inference for signal recovery in non-gaussian regimes,

    M. A. Mousa, L. Bauer, Z. Yang, U. Singh, A. Deka, and Z. Jacob, “Physics-integrated inference for signal recovery in non-gaussian regimes,”Applied Physics Letters, vol. 128, no. 17, p. 171101, 2026. [Online]. Available: https://doi.org/10.1063/5.0324166

    work page doi:10.1063/5.0324166 2026

  42. [42]

    Pid: Physics-informed diffusion model for infrared image generation,

    F. Mao, J. Mei, S. Lu, F. Liu, L. Chen, F. Zhao, and Y . Hu, “Pid: Physics-informed diffusion model for infrared image generation,”Pat- tern Recognition, vol. 169, p. 111816, 2026

    2026

  43. [43]

    Thermal-physics-informed 3d gaussian splatting for infrared images rendering,

    Y . Zhao, H. Sun, C. Liu, and Z. Qi, “Thermal-physics-informed 3d gaussian splatting for infrared images rendering,”Infrared Physics & Technology, p. 106593, 2026

    2026

  44. [44]

    Pcmamba: Physics-informed cross-modal state space model for dual-camera compressive hyperspectral imaging,

    G. Meng, Z. Cai, J. Tu, Y . Wang, C. Li, Y . Huang, and X. Ding, “Pcmamba: Physics-informed cross-modal state space model for dual-camera compressive hyperspectral imaging,”arXiv preprint arXiv:2505.16373, 2025

    work page arXiv 2025

  45. [45]

    Convergence of a block coordinate descent method for nondifferentiable minimization,

    P. Tseng, “Convergence of a block coordinate descent method for nondifferentiable minimization,”Journal of Optimization Theory and Applications, vol. 109, no. 3, pp. 475–494, 2001

    2001

  46. [46]

    Hyperspectral subspace identification,

    J. M. Bioucas-Dias and J. M. P. Nascimento, “Hyperspectral subspace identification,”IEEE Transactions on Geoscience and Remote Sensing, vol. 46, no. 8, pp. 2435–2445, 2008

    2008

  47. [47]

    Fast hyperspectral image denoising and inpainting based on low-rank and sparse representations,

    L. Zhuang and J. M. Bioucas-Dias, “Fast hyperspectral image denoising and inpainting based on low-rank and sparse representations,”IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, vol. 11, no. 3, pp. 730–742, 2018

    2018

  48. [48]

    The libradtran software package for radiative transfer calculations (version 2.0.1),

    C. Emde, R. Buras-Schnell, A. Kylling, B. Mayer, J. Gasteiger, U. Hamann, J. Kylling, B. Richter, C. Pause, T. Dowlinget al., “The libradtran software package for radiative transfer calculations (version 2.0.1),”Geoscientific Model Development, vol. 9, no. 5, pp. 1647–1672, 2016

    2016

  49. [49]

    Baseline correction with asymmetric least squares smoothing,

    P. H. C. Eilers and H. F. M. Boelens, “Baseline correction with asymmetric least squares smoothing,”Leiden University Medical Centre Report, vol. 1, no. 1, p. 5, 2005

    2005