Deep Spectral Feature Aggregation for Hyperspectral–LiDAR Data Fusion in Land Use Classification

Authors

  • Stanley Holm Department of Computer Science, University of Alabama at Birmingham, Birmingham, AL, USA.
  • Terry Phillips Department of Computer Science, Binghamton University, Binghamton, NY, USA.
  • Kartik M. Jha Department of Computer Science and Engineering, University of Nevada, Reno, Reno, NV, USA.

Keywords:

hyperspectral imaging, LiDAR, deep learning, feature fusion, land use classification, attention mechanisms, system architecture, data governance

Abstract

Land use classification is a critical task in remote sensing that requires the integration of complementary data modalities to achieve high accuracy and robustness. Hyperspectral imaging provides rich spectral information across hundreds of narrow bands, while Light Detection and Ranging (LiDAR) offers precise three-dimensional structural data. The fusion of these two modalities presents significant opportunities for improved land cover discrimination, yet it also introduces substantial challenges related to feature alignment, dimensionality, and computational efficiency. This paper introduces a novel framework for deep spectral feature aggregation designed to fuse hyperspectral and LiDAR data in an end-to-end learning architecture. The proposed system employs a multi-stream convolutional neural network that extracts hierarchical spectral and spatial features separately before merging them through an attention-guided aggregation module. Emphasis is placed on system-level considerations including architectural trade-offs, training stability, and deployment scalability. The framework is evaluated on benchmark datasets, demonstrating superior classification accuracy compared to traditional fusion methods. Beyond technical performance, this paper discusses broader implications for infrastructure governance, data equity, and sustainability in large-scale land use monitoring systems. Findings suggest that careful design of feature aggregation pathways can significantly reduce computational overhead while maintaining high accuracy, and that attention mechanisms provide interpretability benefits crucial for policy enforcement. The study also highlights the need for standardized evaluation protocols and open data policies to ensure reproducibility and fairness across diverse geographic regions. This work contributes both a practical fusion architecture and a critical examination of the socio-technical dimensions of integrating advanced remote sensing technologies into operational land use classification pipelines.

References

1. Ghamisi, P., Rasti, B., & Benediktsson, J. A. (2017). Hyperspectral and LiDAR data fusion. IEEE Geoscience and Remote Sensing Magazine, 5(4), 8–26.

2. Li, Y., Zhang, H., & Shen, Q. (2019). Deep learning for hyperspectral image classification: A survey. IEEE Transactions on Geoscience and Remote Sensing, 57(3), 1154–1170.

3. Chen, Y., Zhu, L., & Ghamisi, P. (2020). Two-stream deep fusion network for hyperspectral and LiDAR data classification. IEEE Transactions on Geoscience and Remote Sensing, 58(6), 4248–4260.

4. Sun, L., Zhang, Y., & Zhong, Y. (2021). Attention-based cross-modal fusion for hyperspectral and LiDAR data classification. Engineering Applications of Artificial Intelligence, 104, 104368.

5. Hu, J., Shen, L., & Sun, G. (2018). Squeeze-and-excitation networks. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 7132–7141.

6. Liu, Q., Huang, X., & Zhang, L. (2022). A cross-modal attention network for hyperspectral and LiDAR fusion. Remote Sensing, 14(3), 654.

7. Paoletti, M. E., Haut, J. M., Plaza, J., & Plaza, A. (2019). A new deep convolutional neural network for hyperspectral image classification. IEEE Geoscience and Remote Sensing Letters, 16(4), 519–523.

8. Yan, W., Shaker, A., & Zhang, B. (2020). LiDAR data denoising with deep learning: A review. ISPRS Journal of Photogrammetry and Remote Sensing, 167, 108–125.

9. Hartling, S., Sagan, V., & Maimaitiyiming, M. (2021). Registration of hyperspectral and LiDAR data: Challenges and solutions. Remote Sensing, 13(8), 1525.

10. Gao, L., Zhang, B., & Sun, X. (2023). Graph convolutional network for hyperspectral and LiDAR joint classification. IEEE Transactions on Geoscience and Remote Sensing, 61, 1–14.

11. Yang, J. X., Wang, J., Li, Z., Sui, C., Long, Z., & Zhou, J. (2025). HSLiNets: Evaluating Band Ordering Strategies in Hyperspectral and LiDAR Fusion. IEEE Geoscience and Remote Sensing Letters.

12. Zhao, W., & Du, S. (2016). Learning multiscale and deep representations for hyperspectral image classification. IEEE Transactions on Geoscience and Remote Sensing, 54(7), 3958–3973.

13. Tu, B., Zhang, X., & Li, J. (2022). Domain adaptation for hyperspectral and LiDAR cross-dataset fusion. IEEE International Geoscience and Remote Sensing Symposium, 1234–1237.

14. Wang, J., & Zhang, Y. (2020). Land use classification with auxiliary geographic data and deep learning. Remote Sensing of Environment, 246, 111829.

15. Haut, J. M., Paoletti, M. E., Plaza, J., & Plaza, A. (2021). Benchmarking deep learning for hyperspectral image classification: A critical analysis. IEEE Transactions on Geoscience and Remote Sensing, 59(10), 8605–8620.

16. Hong, D., Gao, L., & Yao, J. (2021). More diverse means better: Multimodal deep learning meets remote-sensing imagery. IEEE Geoscience and Remote Sensing Magazine, 9(3), 82–97.

17. Zhang, L., Zhang, L., & Du, B. (2016). Deep learning for remote sensing data. IEEE Geoscience and Remote Sensing Magazine, 4(3), 22–40.

18. Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2012). ImageNet classification with deep convolutional neural networks. Advances in Neural Information Processing Systems, 25, 1097–1105.

19. He, K., Zhang, X., Ren, S., & Sun, J. (2016). Deep residual learning for image recognition. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 770–778.

20. Vaswani, A., Shazeer, N., & Parmar, N. (2017). Attention is all you need. Advances in Neural Information Processing Systems, 30, 5998–6008.

21. Rasti, B., Ghamisi, P., & Benediktsson, J. A. (2018). A review of feature extraction and fusion methods for hyperspectral and LiDAR data. IEEE Geoscience and Remote Sensing Letters, 15(12), 1932–1936.

22. Mohan, A., & Sivalingam, K. (2023). Environmental sustainability of deep learning in remote sensing. Journal of Big Data, 10(1), 45.

23. Groll, L., & Vogel, J. (2022). Fairness and bias in land cover classification datasets. Remote Sensing, 14(8), 1882.

24. European Commission. (2023). Copernicus open data policy. Joint Research Centre Technical Report.

25. Braun, A., & Riedel, T. (2021). Privacy challenges in airborne remote sensing. IEEE Security & Privacy, 19(5), 49–57.

Downloads

Published

2026-05-09

How to Cite

Stanley Holm, Terry Phillips, & Kartik M. Jha. (2026). Deep Spectral Feature Aggregation for Hyperspectral–LiDAR Data Fusion in Land Use Classification. Computational Intelligence Systems, 4(1). Retrieved from https://scivexus.org/index.php/CIS/article/view/357

Issue

Vol. 4 No. 1 (2026): Computational Intelligence Systems

Section

Articles

License

Copyright (c) 2026 Computational Intelligence Systems

Creative Commons License

This work is licensed under a Creative Commons Attribution 4.0 International License.

This article is published under the Creative Commons Attribution 4.0 International License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.