Cross-Domain Classification for Hyperspectral Imagery: A Comprehensive Review of Domain Adaptation, Domain Generalization, and Multimodal Approaches
DOI:
https://doi.org/10.54097/c0s57j81Keywords:
Hyperspectral image classification, domain adaptation, domain generalization, cross-scene classification, transfer learning, contrastive learning, graph neural networks, multimodal learningAbstract
Cross-domain hyperspectral image (HSI) classification has emerged as a critical challenge in remote sensing, driven by the inherent spectral variability across different acquisition conditions, sensors, and geographical regions. Traditional supervised classification methods rely heavily on abundant labeled data, which is expensive and time-consuming to obtain for each new scene. This review provides a comprehensive synthesis of 23 representative works published between 2015 and 2025, organized into three methodological paradigms: (1) domain adaptation (DA)—13 methods that align source and target distributions through statistical matching, adversarial learning, contrastive learning, or active sample selection; (2) domain generalization (DG)—5 methods that learn domain-invariant representations without target data exposure; and (3) cross-modality and graph-based approaches—5 methods that leverage multimodal data or capture nonlocal topological structures. Our cross-cutting synthesis yields four key findings: (i) feature-level alignment alone is insufficient—class-conditional, decision-boundary, and topological alignment each provide complementary signals; (ii) contrastive learning has emerged as a unifying paradigm since 2023, adopted by 5 of the 10 most recent works; (iii) only 2 of 23 methods support open-set transfer, revealing a critical gap between research assumptions and real-world deployment; and (iv) domain generalization methods have narrowed the accuracy gap with DA while offering zero-shot deployment capability. We identify seven research gaps, each assessed for urgency and impact, with foundation model adaptation, benchmark standardization, and open-set domain adaptation rated as near-term, high-impact priorities. This review provides researchers with a structured taxonomy, a 23-method comparative matrix across six design dimensions, and a prioritized roadmap for future investigations in cross-domain remote sensing image classification.
Downloads
References
[1] Deng, C., Liu, X., Li, C., & Tao, D. (2018). Active multi-kernel domain adaptation for hyperspectral image classification. Pattern Recognition, 77, 306–315. https://doi.org/10.1016/j.patcog.2017.12.006
[2] Ding, Y., Chong, Y., Pan, S., Wang, Y., & Nie, C. (2023). Spatial–spectral unified adaptive probability graph convolutional networks for hyperspectral image classification. IEEE Transactions on Neural Networks and Learning Systems, 34(7), 3653–3667. https://doi.org/10.1109/TNNLS.2022.3141286
[3] Gao, Z., Pan, B., Xu, T., Li, T., & Shi, Z. (2023). LiCa: Label-indicate-conditional-alignment domain generalization for pixel-wise hyperspectral imagery classification. IEEE Transactions on Geoscience and Remote Sensing, 61, 5513516. https://doi.org/10.1109/TGRS.2023.3284098
[4] Hong, D., Yokoya, N., Ge, N., Chanussot, J., & Zhu, X. X. (2019). Learnable manifold alignment (LeMA): A semi-supervised cross-modality learning framework for land cover and land use classification. ISPRS Journal of Photogrammetry and Remote Sensing, 147, 193–205. https://doi.org/10.1016/j.isprsjprs.2018.11.014
[5] Hong, D., Yao, J., Meng, D., Xu, Z., & Chanussot, J. (2021). Multimodal GANs: Toward crossmodal hyperspectral–multispectral image segmentation. IEEE Transactions on Geoscience and Remote Sensing, 59(6), 5103–5113. https://doi.org/10.1109/TGRS.2020.3032203
[6] Huang, Y., Peng, J., Sun, W., Chen, N., Du, Q., Ning, Y., & Su, H. (2022). Two-branch attention adversarial domain adaptation network for hyperspectral image classification. IEEE Transactions on Geoscience and Remote Sensing, 60, 5536914. https://doi.org/10.1109/TGRS.2022.3188944
[7] Kutbi, M., Peng, K.-C., & Wu, Z. (2021). Zero-shot deep domain adaptation with common representation learning. IEEE Transactions on Pattern Analysis and Machine Intelligence, 44(9), 5430–5446. https://doi.org/10.1109/TPAMI.2021.3060090
[8] Li, Z., Xu, Q., Ma, L., Fang, Z., Wang, Y., He, W., & Du, Q. (2023). Supervised contrastive learning-based unsupervised domain adaptation for hyperspectral image classification. IEEE Transactions on Geoscience and Remote Sensing, 61, 5528617. https://doi.org/10.1109/TGRS.2023.3296323
[9] Lin, J., Zhao, L., Li, S., Ward, R., & Wang, Z. J. (2018). Active-learning-incorporated deep transfer learning for hyperspectral image classification. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 11(11), 4249–4262. https://doi.org/10.1109/JSTARS.2018.2868438
[10] Liu, Z., Ma, L., & Du, Q. (2021). Class-wise distribution adaptation for unsupervised classification of hyperspectral remote sensing images. IEEE Transactions on Geoscience and Remote Sensing, 59(1), 508–521. https://doi.org/10.1109/TGRS.2020.2994443
[11] Matasci, G., Volpi, M., Kanevski, M., Bruzzone, L., & Tuia, D. (2015). Semisupervised transfer component analysis for domain adaptation in remote sensing image classification. IEEE Transactions on Geoscience and Remote Sensing, 53(7), 3550–3564. https://doi.org/10.1109/TGRS.2014.2380378
[12] Ning, Y., Peng, J., Liu, Q., Huang, Y., Sun, W., & Du, Q. (2023). Contrastive learning based on category matching for domain adaptation in hyperspectral image classification. IEEE Transactions on Geoscience and Remote Sensing, 61, 5522915. https://doi.org/10.1109/TGRS.2023.3269127
[13] Peng, J., Sun, W., Ma, L., & Du, Q. (2019). Discriminative transfer joint matching for domain adaptation in hyperspectral image classification. IEEE Geoscience and Remote Sensing Letters, 16(6), 953–957. https://doi.org/10.1109/LGRS.2019.2893141
[14] Qin, B., Feng, S., Zhao, C., Xi, B., Li, W., & Tao, R. (2025). FDGNet: Frequency disentanglement and data geometry for domain generalization in cross-scene hyperspectral image classification. IEEE Transactions on Neural Networks and Learning Systems, 36(5), 7171–7185. https://doi.org/10.1109/TNNLS.2024.3484707
[15] Yao, J., Zhang, B., Li, C., Hong, D., & Chanussot, J. (2023). Extended vision transformer (ExViT) for land use and land cover classification: A multimodal deep learning framework. IEEE Transactions on Geoscience and Remote Sensing, 61, 5517915. https://doi.org/10.1109/TGRS.2023.3260295
[16] Zhang, Y., Li, W., Tao, R., Peng, J., Du, Q., & Cai, Z. (2021). Cross-scene hyperspectral image classification with discriminative cooperative alignment. IEEE Transactions on Geoscience and Remote Sensing, 59(9), 7646–7660. https://doi.org/10.1109/TGRS.2021.3060137
[17] Zhang, Y., Li, W., Sun, W., Tao, R., & Du, Q. (2023). Single-source domain expansion network for cross-scene hyperspectral image classification. IEEE Transactions on Image Processing, 32, 1499–1514. https://doi.org/10.1109/TIP.2023.3241811
[18] Zhang, Y., Zhang, M., Li, W., Wang, S., & Tao, R. (2023). Language-aware domain generalization network for cross-scene hyperspectral image classification. IEEE Transactions on Geoscience and Remote Sensing, 61, 5502812. https://doi.org/10.1109/TGRS.2023.3244601
[19] Zhang, Y., Li, W., Zhang, M., Qu, Y., Tao, R., & Qi, H. (2023). Topological structure and semantic information transfer network for cross-scene hyperspectral image classification. IEEE Transactions on Neural Networks and Learning Systems, 34(6), 3112–3125. https://doi.org/10.1109/TNNLS.2022.3141964
[20] Zhang, Y., Li, W., Zhang, M., Wang, S., Tao, R., & Du, Q. (2024). Graph information aggregation cross-domain few-shot learning for hyperspectral image classification. IEEE Transactions on Neural Networks and Learning Systems, 35(4), 5712–5726. https://doi.org/10.1109/TNNLS.2023.3314276
[21] Zhao, C., Qin, B., Feng, S., Zhu, W., Zhang, L., & Ren, J. (2022). An unsupervised domain adaptation method towards multi-level features and decision boundaries for cross-scene hyperspectral image classification. IEEE Transactions on Geoscience and Remote Sensing, 60, 5545916. https://doi.org/10.1109/TGRS.2022.3189122
[22] Zhao, H., Zhang, J., Lin, L., Wang, J., Gao, S., & Zhang, Z. (2023). Locally linear unbiased randomization network for cross-scene hyperspectral image classification. IEEE Transactions on Geoscience and Remote Sensing, 61, 5527215. https://doi.org/10.1109/TGRS.2023.3294820
[23] Zhu, Y., Zhuang, F., Wang, J., Chen, Z., Shi, Z., & Wu, W. (2019). Multi-representation adaptation network for cross-domain image classification. Neural Networks, 119, 214–221. https://doi.org/10.1016/j.neunet.2019.08.011
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Academic Journal of Applied Sciences
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
