pith. sign in

arxiv: 2606.07985 · v1 · pith:BU7S6LNGnew · submitted 2026-06-06 · 💻 cs.CV · cs.CL

FMRFusion: Frequency-Aware Multi-View Representation Learning for Heterogeneous Image Fusion

Tao Zhoua , Yunlong Liu , Qinghui Chen , Zekai Zhang , Minlong Sun , Changlin Biana , Dagang Li , Wenmin Wang
show 1 more author
Jinglin Zhang
This is my paper

Pith reviewed 2026-06-27 20:17 UTC · model grok-4.3

classification 💻 cs.CV cs.CL
keywords infrared visible image fusionmulti-view representation learningfrequency decompositioncross-view interactionflow matchingheterogeneous image fusionnighttime fusion
0
0 comments X

The pith

FMRFusion fuses infrared and visible images by separating frequencies and modeling cross-view complementary features instead of stacking single modules.

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

The paper introduces FMRFusion to fuse infrared and visible images into one composite that keeps both target details and textures. It argues that earlier methods stack the same module on both modalities and therefore miss their distinct traits, especially under real-world variation. The new design adds a multi-scale structure module, splits features into high- and low-frequency parts with bilinear decomposition, lets the two views interact explicitly, and refines the result with flow matching. These steps are shown to raise performance on standard benchmarks, with the largest gains appearing in nighttime scenes. A reader would care because many vision systems rely on reliable fusion when one sensor sees what the other cannot.

Core claim

The central claim is that a frequency-aware multi-view network overcomes the incomplete feature learning of prior single-module stacking methods. The Multi-Scale Structural Perception Module extracts local and contextual structures, bilinear frequency decomposition jointly models high-frequency details and low-frequency global content, Cross-View Complementary Interaction fuses reflected-light and radiative-intensity information, and flow matching progressively improves the fused representation from coarse to high-quality output. Experiments on multiple datasets show this combination yields superior and consistent fusion results, particularly in nighttime conditions.

What carries the argument

The FMRFusion architecture built around the Multi-Scale Structural Perception Module, bilinear frequency decomposition, Cross-View Complementary Interaction module, and flow-matching refinement.

If this is right

  • Fused images retain more target information from the infrared modality while preserving texture from the visible modality.
  • Performance stays higher across varied real-world heterogeneous data than single-module baselines.
  • Nighttime fusion quality improves noticeably because radiative and reflective cues are modeled together.
  • Flow matching supplies a progressive refinement step that can be applied after initial feature fusion.

Where Pith is reading between the lines

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

  • The same frequency-split and cross-view pattern could be tested on other paired sensor data such as RGB-depth or medical modalities.
  • If the bilinear decomposition proves stable, it might replace hand-crafted frequency filters in other detail-preserving vision pipelines.
  • Flow matching as a post-fusion step suggests a route to iterative quality improvement without retraining the entire encoder.

Load-bearing premise

Single-module stacking leaves the distinct characteristics of infrared and visible modalities incompletely learned, and the proposed multi-view modules will fix that gap.

What would settle it

An ablation that removes the frequency decomposition and cross-view interaction modules and measures whether fusion metrics on nighttime test sets remain unchanged or improve.

Figures

Figures reproduced from arXiv: 2606.07985 by Changlin Biana, Dagang Li, Jinglin Zhang, Minlong Sun, Qinghui Chen, Tao Zhoua, Wenmin Wang, Yunlong Liu, Zekai Zhang.

Figure 1
Figure 1. Figure 1: Illustration of FMRFusion. In the feature decomposition stage, the infrared and visible images pairs I, V are fed into the shallow DRSformer encoder SDE to obtain the features F S I ,F S V , respectively. The feature F S I is inputted into the Multi-Scale Structural Perception Module (MS-SPM) and Transformer Block (TRB) for feature decomposition, with the same process applied to F S V . It is worth mention… view at source ↗
Figure 2
Figure 2. Figure 2: Illustration of core modules in FMRFusion. The proposed framework consists of the EAB, GAB, and CVCI. The EAB enhances feature representation, the GAB strengthens structural details via gradient information, and the CVCI enables cross-modal interaction through cross￾attention. 5Additionally, intensity enhancement and color denoising layers are employed to improve the quality of fused features. 3.1.2. Fusio… view at source ↗
Figure 3
Figure 3. Figure 3: Flow-Matching-Based Fusion Framework for Infrared–Visible Image Enhancement.Illustration of the Flow-Matching-based enhancement module applied to infrared–visible image fusion. The module progressively refines coarse Stage-II fusion results under semantic guidance using a conditional U-Net to estimate velocity fields. Key components include semantic masks generated from input images, linear interpolation t… view at source ↗
Figure 4
Figure 4. Figure 4: Visual comparison for “01364N” in MSRS dataset [PITH_FULL_IMAGE:figures/full_fig_p011_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Visual comparison of “2” from the TNO dataset. Second, regarding the TNO dataset, the proposed method delivers the best performance across the six evaluation metrics, demonstrating strong overall effectiveness. In particular, it obtains the highest scores on EN, SD, and SF, indicating that the fused images contain richer information content, higher contrast, and more abundant texture details. Moreover, our… view at source ↗
Figure 6
Figure 6. Figure 6: Visual comparison for “FLIR 05857” in RoadScene dataset [PITH_FULL_IMAGE:figures/full_fig_p012_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Visual comparison for “00449” in M3FD dataset. Third, for the RoadScene dataset, our method achieves the best performance across all evaluation metrics, demon￾strating clear overall superiority. Specifically, it obtains the highest scores on SF and AG, indicating richer texture details and clearer structural information in the fused images. Our method also ranks first on SD, showing its strong ability to e… view at source ↗
Figure 8
Figure 8. Figure 8: Visual comparison for“MRI-PET-28”in MRI-PET dataset 6. Multi-Focus Image Fusion In order to verify the generalizability of the model, we tested it directly using 639 pairs of images from the RealMFF dataset as a training target for the multifocus task. Multi-focus image fusion was tested using 71 pairs of images from RealMFF, 20 pairs of images from Lytro to validate its effectiveness, and the fusion resul… view at source ↗
Figure 9
Figure 9. Figure 9: Visual comparison of fusion results in multifocal image fusion tasks. In the multi-focus image fusion experiments on Lytro dataset as shown [PITH_FULL_IMAGE:figures/full_fig_p016_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Visual results of the object detection task on the MSRS dataset [PITH_FULL_IMAGE:figures/full_fig_p017_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Visual results of the object detection task on the MSRS dataset. 8. Conclusion In this work, we propose FMRFusion, a frequency-aware multi-view representation learning network that targets several critical challenges in heterogeneous data fusion. First, insufficient decoupling of view-specific information is addressed through explicit high-/low-frequency decomposition, which alleviates the loss of structu… view at source ↗
read the original abstract

Infrared and visible image fusion aims to generate a composite image that retains significant target information and preserves detailed textures, integrating two heterogeneous modalities. Previous image fusion methods typically adopt a single-module stacking approach to extract features from the two modalities. However, these approaches may result in incomplete learning of their distinct characteristics, thereby limiting the fusion effectiveness and constrain ing robustness in real-world heterogeneous data scenarios. To address these challenges, we propose FMRFusion, a frequency-aware multi-view representation learning network for Heterogeneous Image Fusion. A Multi-Scale Struc tural Perception Module is introduced to effectively capture discriminative structures, extracting fine-grained local structures and essential contextual information. A bilinear frequency decomposition mechanism is employed to sepa rate features into high-frequency and low-frequency components, enabling joint modeling of local details and global representations across different frequency domains. Moreover, a Cross-View Complementary Interaction is incorpo rated to explicitly model and fuse the complementary characteristics between reflected light information and radiative intensity responses, facilitating effective cross-view interaction. We further improve the Performance of the fused results by flow matching, which progressively refines the fused features by learning the transformation from coarse data to high-quality representations. Extensive experiments conducted on multiple benchmark datasets demonstrate that FMRFusion achieves superior and consistent performance across a range of fusion tasks, especially in nighttime scenarios

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 / 0 minor

Summary. The paper proposes FMRFusion, a frequency-aware multi-view representation learning network for infrared and visible image fusion. It introduces a Multi-Scale Structural Perception Module to capture discriminative structures, a bilinear frequency decomposition mechanism to separate high- and low-frequency components, a Cross-View Complementary Interaction module to model complementary characteristics between modalities, and flow matching to refine fused features. The central claim is that this architecture overcomes incomplete modality learning from prior single-module stacking approaches and achieves superior, consistent performance on benchmark datasets, especially in nighttime scenarios.

Significance. If the experimental claims hold, the explicit multi-view and frequency-aware design could advance heterogeneous fusion by better preserving both structural details and global context across modalities, with potential benefits for nighttime robustness in applications such as surveillance and autonomous navigation.

major comments (1)
  1. Abstract: performance claims of 'superior and consistent performance across a range of fusion tasks' are asserted without any quantitative results, baselines, error bars, dataset names, or experimental protocol, rendering the central claim that the proposed modules overcome single-module limitations impossible to evaluate from the provided text.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive comment on the abstract. We agree that incorporating specific quantitative highlights will strengthen the presentation of our claims while maintaining the abstract's conciseness.

read point-by-point responses

  1. Referee: [—] Abstract: performance claims of 'superior and consistent performance across a range of fusion tasks' are asserted without any quantitative results, baselines, error bars, dataset names, or experimental protocol, rendering the central claim that the proposed modules overcome single-module limitations impossible to evaluate from the provided text.

    Authors: The abstract is a high-level summary, with full quantitative support (including named datasets such as TNO and RoadScene, multiple baselines, metrics like PSNR/SSIM, and standard deviations) provided in Section 4 of the manuscript. We will revise the abstract to include 1-2 key quantitative results (e.g., average gains in nighttime scenarios) to directly address this point. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper proposes an architectural framework (multi-scale structural perception, bilinear frequency decomposition, cross-view interaction, flow matching) to address limitations of prior single-module stacking methods. No equations, derivations, fitted parameters renamed as predictions, or self-citation chains appear in the provided abstract or description. The central claims rest on empirical benchmark results rather than any self-referential reduction of outputs to inputs by construction. This is a standard empirical ML architecture paper with no load-bearing mathematical steps that collapse into tautology.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract contains no mathematical derivations, free parameters, axioms, or postulated entities; all assessments are limited by the absence of the full manuscript.

pith-pipeline@v0.9.1-grok · 5789 in / 1019 out tokens · 30235 ms · 2026-06-27T20:17:29.501597+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

68 extracted references · 4 canonical work pages

  1. [1]

    H. Xu, R. Nie, J. Cao, G. Xie, Z. Ding, Imqfusion: Infrared and visible image fusion via implicit multi-resolution preservation and query aggregation, Expert Systems with Applications 257 (2024) 125014

    2024

  2. [2]

    J. Liu, G. Wu, Z. Liu, D. Wang, Z. Jiang, L. Ma, W. Zhong, X. Fan, R. Liu, Infrared and visible image fusion: From data compatibility to task adaption, IEEE Transactions on Pattern Analysis and Machine Intelligence 47 (2024) 2349–2369

    2024

  3. [3]

    H. Zhao, T. Jiang, X. Li, J. Jin, Sfdfuse: A lightweight spatial-frequency fusion network for infrared-visible images, Pattern Recognition (2026) 113634

    2026

  4. [4]

    H. Li, Z. Yang, Y . Zhang, W. Jia, Z. Yu, Y . Liu, Mulfs-cap: Multimodal fusion-supervised cross-modality alignment perception for unregistered infrared-visible image fusion, IEEE Transactions on Pattern Analysis and Machine Intelligence 47 (2025) 3673–3690. doi:10.1109/TPAMI.2025.3535617

    work page doi:10.1109/tpami.2025.3535617 2025

  5. [5]

    J. Dong, W. Wu, J. Cheng, X. Tang, You sense only once beneath: Ultra-light real-time underwater object detection, arXiv preprint arXiv:2504.15694 (2025)

    arXiv 2025

  6. [6]

    Qiang, Y

    Z. Qiang, Y . Shen, Y . Yuan, G. Pei, Dwsfusion: Dual weight supervision for lightweight infrared and visible image fusion, Pattern Recognition (2026) 113520

    2026

  7. [7]

    B. Tang, Z. Yang, D. Yao, B. Yang, L. Chen, R. Bi, Y . Zhao, J. Xie, Z. Guo, Loft-clip: Few-shot anomaly detection for railway fasteners based on large vision-language models, Pattern Recognition 180 (2026) 114099. doi:10.1016/j.patcog.2026.114099

    work page doi:10.1016/j.patcog.2026.114099 2026

  8. [8]

    Zhang, Z

    Y . Zhang, Z. Wang, M. Huang, M. Li, J. Zhang, S. Wang, J. Zhang, H. Zhang, S2dbft: Spectral-spatial dual- branch fusion transformer for hyperspectral image classification, IEEE Transactions on Geoscience and Remote Sensing (2025). 18

    2025

  9. [9]

    Zhang, J.-F

    J. Zhang, J.-F. Nezan, J.-G. Cousin, Implementation of motion estimation based on heterogeneous parallel computing system with opencl, in: 2012 IEEE 14th International Conference on High Performance Computing and Communication & 2012 IEEE 9th International Conference on Embedded Software and Systems, IEEE, 2012, pp. 41–45

    2012

  10. [10]

    Y . Feng, J. Zheng, M. Qin, C. Bai, J. Zhang, 3d octave and 2d vanilla mixed convolutional neural network for hyperspectral image classification with limited samples, Remote Sensing 13 (2021) 4407

    2021

  11. [11]

    M. Gao, X. He, L. Chen, T. Liu, J. Zhang, A. Zhou, Learning vertex representations for bipartite networks, IEEE transactions on knowledge and data engineering 34 (2020) 379–393

    2020

  12. [12]

    P. Zhu, Z. Zhu, Y . Wang, J. Zhang, S. Zhao, Multi-granularity episodic contrastive learning for few-shot learning, Pattern Recognition 131 (2022) 108820

    2022

  13. [13]

    Z. Zhou, F. Zhang, H. Xiao, F. Wang, X. Hong, K. Wu, J. Zhang, A novel ground-based cloud image seg- mentation method by using deep transfer learning, IEEE Geoscience and Remote Sensing Letters 19 (2021) 1–5

    2021

  14. [14]

    Y . Jia, D. Yao, J. Yang, L. Jia, T. Zhou, Q. Long, A multi-scale wavelet and spectral kurtosis-guided frame- work for noise-resistant rolling bearing fault diagnosis, Mechanical Systems and Signal Processing 242 (2026) 113662

    2026

  15. [15]

    J. Chen, W. Yu, Z. Cheng, X. Tian, J. Ma, Tofusion: Text-guided and object-aware infrared and visible image fusion, Pattern Recognition (2026) 113569

    2026

  16. [16]

    J. Ma, L. Tang, F. Fan, J. Huang, X. Mei, Y . Ma, Swinfusion: Cross-domain long-range learning for general image fusion via swin transformer, IEEE/CAA Journal of Automatica Sinica 9 (2022) 1200–1217

    2022

  17. [17]

    Zhang, J

    H. Zhang, J. Ma, Sdnet: A versatile squeeze-and-decomposition network for real-time image fusion, Interna- tional Journal of Computer Vision 129 (2021) 2761–2785

    2021

  18. [18]

    Zhang, Y

    Y . Zhang, Y . Liu, P. Sun, H. Yan, X. Zhao, L. Zhang, Ifcnn: A general image fusion framework based on convolutional neural network, Information Fusion 54 (2020) 99–118

    2020

  19. [19]

    L. Qu, S. Liu, M. Wang, Z. Song, Transmef: A transformer-based multi-exposure image fusion framework using self-supervised multi-task learning, in: Proceedings of the AAAI conference on artificial intelligence, volume 36, 2022, pp. 2126–2134

    2022

  20. [20]

    H. Xu, J. Ma, J. Yuan, Z. Le, W. Liu, Rfnet: Unsupervised network for mutually reinforcing multi-modal image registration and fusion, in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 19679–19688

    2022

  21. [21]

    Y . Long, H. Jia, Y . Zhong, Y . Jiang, Y . Jia, Rxdnfuse: A aggregated residual dense network for infrared and visible image fusion, Information Fusion 69 (2021) 128–141

    2021

  22. [22]

    Y . Wang, X. Zhao, J. Pu, L. Zhang, D. Miao, Multi-scale frequency attention fusion network for infrared and visible image fusion, Engineering Applications of Artificial Intelligence 159 (2025) 111728

    2025

  23. [23]

    C. Li, R. Zhu, H. Zhao, X. Li, X. Zhang, Atdfusion: Adapter-tuned dual-branch network for multimodal medical image fusion, Pattern Recognition (2026) 113848

    2026

  24. [24]

    Zhang, P

    J. Zhang, P. Liu, F. Zhang, H. Iwabuchi, A. A. d. H. e Ayres, V . H. C. De Albuquerque, et al., Ensemble meteorological cloud classification meets internet of dependable and controllable things, IEEE Internet of Things Journal 8 (2020) 3323–3330

    2020

  25. [25]

    M. Miao, W. Hu, B. Xu, J. Zhang, J. J. Rodrigues, V . H. C. De Albuquerque, Automated cca-mwf algorithm for unsupervised identification and removal of eog artifacts from eeg, IEEE Journal of Biomedical and Health Informatics 26 (2021) 3607–3617. 19

    2021

  26. [26]

    Q. Ma, C. Bai, J. Zhang, Z. Liu, S. Chen, Supervised learning based discrete hashing for image retrieval, Pattern Recognition 92 (2019) 156–164

    2019

  27. [27]

    Y . Li, Y . Yang, K. Zhu, J. Zhang, Clothing sale forecasting by a composite gru–prophet model with an attention mechanism, IEEE Transactions on Industrial Informatics 17 (2021) 8335–8344

    2021

  28. [28]

    Q. Chen, Z. Zhang, Z. Zhang, K. Zhang, D. Li, W. Wang, J. Zhang, C. Liu, Distilled large language model-driven dynamic sparse expert activation mechanism, Applied Soft Computing (2025) 114037

    2025

  29. [29]

    Q. Chen, L. Wang, Z. Zhang, X. Wang, W. Liu, B. Xia, H. Ding, J. Zhang, S. Xu, X. Wang, Dual-path aggregation transformer network for super-resolution with images occlusions and variability, Engineering Applications of Artificial Intelligence 139 (2025)

    2025

  30. [30]

    K. Cui, J. Cheng, Y . Pan, Fdafusion: A joint frequency decomposition and adaptive fusion network for nighttime infrared and visible image fusion, in: 2025 International Joint Conference on Neural Networks (IJCNN), 2025, pp. 1–8. doi:10.1109/IJCNN64981.2025.11229055

    work page doi:10.1109/ijcnn64981.2025.11229055 2025

  31. [31]

    C. Pan, Q. Jiang, H. Zheng, H. Huang, X. Jin, K. Li, W. Zhou, Dmnet: A dense multi-scale feature extraction network with two-stage training for infrared-visible image fusion, IEEE Internet of Things Journal (2025)

    2025

  32. [32]

    H. Xu, J. Ma, J. Jiang, X. Guo, H. Ling, U2fusion: A unified unsupervised image fusion network, IEEE Transactions on Pattern Analysis and Machine Intelligence 44 (2020) 502–518

    2020

  33. [33]

    Li, X.-J

    H. Li, X.-J. Wu, Densefuse: A fusion approach to infrared and visible images, IEEE Transactions on Image Processing 28 (2019) 2614–2623

    2019

  34. [34]

    Li, X.-J

    H. Li, X.-J. Wu, T. Durrani, Nestfuse: An infrared and visible image fusion architecture based on nest connection and spatial/channel attention models, IEEE Transactions on Instrumentation and Measurement 69 (2020) 9645– 9656

    2020

  35. [35]

    Li, X.-J

    H. Li, X.-J. Wu, J. Kittler, Rfn-nest: An end-to-end residual fusion network for infrared and visible images, Information Fusion 73 (2021) 72–86

    2021

  36. [36]

    Y . Guo, R. Xu, R. Li, W. Su, Dae-fuse: An adaptive discriminative autoencoder for multi-modality image fusion, in: 2025 IEEE International Conference on Multimedia and Expo (ICME), IEEE, 2025, pp. 1234–1239

    2025

  37. [37]

    H. Xu, H. Zhang, J. Ma, Classification saliency-based rule for visible and infrared image fusion, IEEE Transac- tions on Computational Imaging 7 (2021) 824–836

    2021

  38. [38]

    Zhang, J

    Z. Zhang, J. Zhou, J. Shi, J. Lu, Cpigan: Infrared and visible image fusion via cross-scale progressive interaction network with adversarial learning, Computers and Electrical Engineering 128 (2025) 110672

    2025

  39. [39]

    J. Ma, H. Zhang, Z. Shao, P. Liang, H. Xu, Ganmcc: A generative adversarial network with multiclassification constraints for infrared and visible image fusion, IEEE Transactions on Instrumentation and Measurement 70 (2020) 1–14

    2020

  40. [40]

    J. Ma, H. Xu, J. Jiang, X. Mei, X.-P. Zhang, Ddcgan: A dual-discriminator conditional generative adversarial network for multi-resolution image fusion, IEEE Transactions on Image Processing 29 (2020) 4980–4995

    2020

  41. [41]

    J. Li, H. Huo, C. Li, R. Wang, Q. Feng, Attentionfgan: Infrared and visible image fusion using attention-based generative adversarial networks, IEEE Transactions on Multimedia 23 (2020) 1383–1396

    2020

  42. [42]

    J. Liu, X. Fan, Z. Huang, G. Wu, R. Liu, W. Zhong, Z. Luo, Target-aware dual adversarial learning and a multi-scenario multi-modality benchmark to fuse infrared and visible for object detection, in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 5802–5811

    2022

  43. [43]

    Z. Ding, R. Hou, Y . Men, S. Luan, Y . Liu, K. He, S. Xie, Dskfuse: Passive-active distillation learning for multi-modal image fusion via dynamic sparse kansformer, Expert Systems with Applications (2025) 130610. 20

    2025

  44. [44]

    Zhang, H

    X. Zhang, H. Qin, J. Geng, J. Liu, Z. Liu, H. Li, Sfdfuse: Spatial and frequency feature decomposition for visible and infrared image fusion, Pattern Recognition (2026) 113949

    2026

  45. [45]

    Q. Chen, Z. Zhang, H. Liu, J. Zhang, C. Bai, Kftd: Koopman-fourier time-differentiable network for continuous ocean spatiotemporal forecasting, in: Proceedings of the 32nd ACM SIGKDD Conference on Knowledge Discovery and Data Mining V . 1, 2026, pp. 94–103

    2026

  46. [46]

    Zhang, G

    Z. Zhang, G. Li, H. Zhang, Q. Chen, Q. Zhang, J. Wan, M. Xiong, C. Bai, D. Li, W. Zhang, et al., A novel dataset and lightweight distillation baseline for highlight transparent object detection, International Journal of Computer Vision 134 (2026) 157

    2026

  47. [47]

    Zhang, M

    Z. Zhang, M. Zhou, H. Wan, M. Li, G. Li, D. Han, Idd-net: Industrial defect detection method based on deep-learning, Engineering Applications of Artificial Intelligence 123 (2023) 106390

    2023

  48. [48]

    Zhang, Q

    Z. Zhang, Q. Chen, M. Xiong, S. Ding, Z. Su, X. Yao, Y . Sun, C. Bai, J. Zhang, Zero-shot learning in industrial scenarios: New large-scale benchmark, challenges and baseline, in: Proceedings of the AAAI Conference on Artificial Intelligence, volume 39, 2025, pp. 10357–10366

    2025

  49. [49]

    Zhang, Z

    J. Zhang, Z. Zhang, Q. Chen, G. Li, W. Li, S. Ding, M. Xiong, W. Zhang, S. Chen, Representation learning based on co-evolutionary combined with probability distribution optimization for precise defect location, IEEE Transactions on Neural Networks and Learning Systems 36 (2024) 11989–12003

    2024

  50. [50]

    Zhang, J

    Z. Zhang, J. Zhang, Q. Chen, G. Li, D. Chen, S. Jing, H. Wang, D. Li, C. Liu, C. Bai, et al., Unification of closed- open industrial detection scenarios: New large-scale benchmarks, challenges and baselines, IEEE Transactions on Pattern Analysis and Machine Intelligence (2026)

    2026

  51. [51]

    H. Yan, S. Xiong, L. Wang, L. Jian, G. Vivone, Atfusion: An alternate cross-attention transformer network for infrared and visible image fusion, Infrared Physics & Technology 152 (2026) 106252

    2026

  52. [52]

    Z. Wang, Y . Chen, W. Shao, H. Li, L. Zhang, Swinfuse: A residual swin transformer fusion network for infrared and visible images, IEEE Transactions on Instrumentation and Measurement 71 (2022) 1–12

    2022

  53. [53]

    W. Tang, F. He, Y . Liu, Y . Duan, T. Si, Datfuse: Infrared and visible image fusion via dual attention transformer, IEEE Transactions on Circuits and Systems for Video Technology (2023)

    2023

  54. [54]

    H. Bai, J. Zhang, Z. Zhao, Y . Wu, L. Deng, Y . Cui, T. Feng, S. Xu, Task-driven image fusion with learnable fusion loss, in: Proceedings of the Computer Vision and Pattern Recognition Conference, 2025, pp. 7457–7468

    2025

  55. [55]

    H. Zhao, R. Nie, Dndt: Infrared and visible image fusion via densenet and dual-transformer, in: 2021 Interna- tional Conference on Information Technology and Biomedical Engineering (ICITBE), IEEE, 2021, pp. 71–75

    2021

  56. [56]

    J. Li, J. Zhu, C. Li, X. Chen, B. Yang, Cgtf: Convolution-guided transformer for infrared and visible image fusion, IEEE Transactions on Instrumentation and Measurement 71 (2022) 1–14

    2022

  57. [57]

    X. Chen, H. Li, M. Li, J. Pan, Learning a sparse transformer network for effective image deraining, in: Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 5896–5905

    2023

  58. [58]

    Y . Jie, Y . Xu, X. Li, H. Tan, Tsjnet: A multi-modality target and semantic awareness joint-driven image fusion network, arXiv preprint arXiv:2402.01212 (2024). https://doi.org/10.1109/tgrs.2023.3243900

    work page doi:10.1109/tgrs.2023.3243900 2024

  59. [59]

    P. Wang, X. Wang, F. Wang, M. Lin, S. Chang, H. Li, R. Jin, Kvt: k-nn attention for boosting vision transformers, in: European conference on computer vision, Springer, 2022, pp. 285–302

    2022

  60. [60]

    Zhang, G

    X. Zhang, G. Fan, Y . Ju, L. Fan, J. Liu, A. C. Kot, et al., Sgdfuse: Sam-guided diffusion for high-fidelity infrared and visible image fusion, arXiv preprint arXiv:2508.05264 (2025)

    arXiv 2025

  61. [61]

    H. Li, T. Xu, X.-J. Wu, J. Lu, J. Kittler, Lrrnet: A novel representation learning guided fusion network for infrared and visible images, IEEE transactions on pattern analysis and machine intelligence (2023). 21

    2023

  62. [62]

    Z. Liu, J. Liu, G. Wu, L. Ma, X. Fan, R. Liu, Bi-level dynamic learning for jointly multi-modality image fusion and beyond, IJCAI (2023)

    2023

  63. [63]

    S. W. Zamir, A. Arora, S. Khan, M. Hayat, F. S. Khan, M.-H. Yang, Restormer: Efficient transformer for high-resolution image restoration, in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 5728–5739

    2022

  64. [64]

    H. Jung, Y . Kim, H. Jang, N. Ha, K. Sohn, Unsupervised deep image fusion with structure tensor representations, IEEE Transactions on Image Processing 29 (2020) 3845–3858

    2020

  65. [65]

    X. Deng, P. L. Dragotti, Deep convolutional neural network for multi-modal image restoration and fusion, IEEE transactions on pattern analysis and machine intelligence 43 (2020) 3333–3348

    2020

  66. [66]

    Liang, J

    P. Liang, J. Jiang, X. Liu, J. Ma, Fusion from decomposition: A self-supervised decomposition approach for image fusion, in: European Conference on Computer Vision, Springer, 2022, pp. 719–735

    2022

  67. [67]

    X. Hu, J. Jiang, X. Liu, J. Ma, Zmff: Zero-shot multi-focus image fusion, Information Fusion 92 (2023) 127–138

    2023

  68. [68]

    Z. Zhao, L. Deng, H. Bai, Y . Cui, Z. Zhang, Y . Zhang, H. Qin, D. Chen, J. Zhang, P. Wang, et al., Image fusion via vision-language model, arXiv preprint arXiv:2402.02235 (2024). 22

    arXiv 2024