EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Self-Supervised Hyperspectral Image Clustering via Spatial–Frequency Interaction and Amplitude–Phase Decoupling

Heng Yuan, Nan Huang, Qichao Liu, Pengfei Liu, Kang Ni, Zhizhong Zheng

Remote Sensing · 31.07.2026 · 18 (15) · 2494

DOI: 10.3390/rs18152494 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Remote Sensing, online erschienen am 31.07.2026. Verfasst von Heng Yuan und 5 weiteren Autor:innen. Die Arbeit nennt 49 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Natural Science Foundation of China, Frontier Technologies R&D Program of Jiangsu, Natural Science Foundation of Jiangsu Higher Education Institutions of China.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Hyperspectral image (HSI) clustering assigns unlabeled pixels to land-cover groups by jointly exploiting spectral and spatial observations. Existing Vision Transformer-based deep clustering captures global dependencies through self-attention. However, the quadratic computational complexity of self-attention restricts practical applications in large HSI scenes. Furthermore, illumination variation and topographic shading shift spectral amplitude of co-class pixels toward divergent directions in feature space, enlarging intra-class distances and reducing inter-class separability in learned embeddings. To address the above limitations, we propose a self-supervised Spatial–Frequency Interaction and Amplitude–Phase Decoupling framework, termed SFI-APD, which integrates a High-Order Spatial–Frequency Interaction Module (HSFIM), a Frequency Feature Attention Block (FFAB), and a Frequency-Domain Vision Transformer (FreqViT) into a unified architecture. Specifically, HSFIM couples local convolutions with Fourier filtering to extract enriched spectral–spatial representations. FFAB then decouples amplitude and phase components to suppress brightness variations, yielding illumination-robust embeddings. Finally, FreqViT performs attention modulation across spectral channels, reducing token aggregation complexity from O(N2D) to O(NDlogN). On the Indian Pines, Salinas, Pavia University, and Yangzhou datasets, SFI-APD achieves OAs of 57.47%, 79.38%, 54.57%, and 64.11%, respectively, outperforming state-of-the-art self-supervised methods for large HSIs.

Förderung

  • National Natural Science Foundation of China
  • Frontier Technologies R&D Program of Jiangsu
  • Natural Science Foundation of Jiangsu Higher Education Institutions of China
  • Deep Earth Probe and Mineral Resources Exploration-National Science and Technology Major Project
  • Jiangsu Geological Bureau Research Project
  • Natural Science Foundation of Jiangsu Province
  • Jiangsu Association for Science and Technology Youth Talent Promotion Program

Zitieren

APA

Yuan, H., Huang, N., Liu, Q., Liu, P., Ni, K., & Zheng, Z. (2026). Self-Supervised Hyperspectral Image Clustering via Spatial–Frequency Interaction and Amplitude–Phase Decoupling. Remote Sensing, 18(15), 2494. https://doi.org/10.3390/rs18152494

Vancouver

Yuan H, Huang N, Liu Q, Liu P, Ni K, Zheng Z. Self-Supervised Hyperspectral Image Clustering via Spatial–Frequency Interaction and Amplitude–Phase Decoupling. Remote Sensing. 2026;18(15):2494. doi:10.3390/rs18152494

BibTeX
@article{Yuan2026,
  author = {Yuan, Heng and Huang, Nan and Liu, Qichao and Liu, Pengfei and Ni, Kang and Zheng, Zhizhong},
  title = {Self-Supervised Hyperspectral Image Clustering via Spatial–Frequency Interaction and Amplitude–Phase Decoupling},
  journal = {Remote Sensing},
  year = {2026},
  volume = {18},
  number = {15},
  pages = {2494},
  doi = {10.3390/rs18152494}
}
Kostenlos starten →