EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

RFM-UNet: Hybrid Frequency–Mamba UNet for Remote-Sensing Road Extraction

Pu Song, Peng Yu, Xiaojing Zhong, Shuizhen Wu, Junbin Tong, Yuanrong He, Lujun Zhang, Guangchun Li, Mengmeng Li

Remote Sensing · 03.08.2026 · 18 (15) · 2546

DOI: 10.3390/rs18152546 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Remote Sensing, online erschienen am 03.08.2026. Verfasst von Pu Song und 8 weiteren Autor:innen. Die Arbeit nennt 59 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Fujian Provincial Natural Science Foundation of China, Natural Science Foundation of Xiamen, Fujian Provincial Science and Technology Innovation Project of Natural Resources.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Road-network extraction from very high-resolution (VHR) remote-sensing imagery remains a challenging task owing to the structural sparsity, topological complexity, and severe occlusions of road networks. Conventional graph-based approaches preserve topological consistency yet incur considerable computational overhead, whereas prevailing convolutional neural network (CNN) and Transformer architectures struggle to reconcile long-range contextual modeling with computational efficiency. To address these limitations, this study proposes RFM-UNet, a hybrid frequency and state–space network designed for road-network segmentation. Specifically, the encoder integrates Mamba blocks with an Anisotropic Directional Attention (ADA) module to jointly capture local geometric cues and global dependencies at linear computational complexity. In addition, a Multi-Scale Adaptive Fusion Module (MAFM) is introduced to dynamically recalibrate multi-stage features, thereby suppressing cross-scale interference and preserving the connectivity of narrow roads. To enhance robustness against shadow-induced occlusions, a Dual-Spectrum Aggregation Module (DualSpec) decouples the phase and amplitude spectra in the frequency domain and fuses them with spatial features, effectively mitigating spurious responses and background noise characterized by similar textures. Quantitative and qualitative experiments on three public datasets demonstrate that RFM-UNet consistently outperforms current state-of-the-art methods.

Förderung

  • Fujian Provincial Natural Science Foundation of China
  • Natural Science Foundation of Xiamen
  • Fujian Provincial Science and Technology Innovation Project of Natural Resources
  • National Natural Science Foundation of China

Zitieren

APA

Song, P., Yu, P., Zhong, X., Wu, S., Tong, J., He, Y., Zhang, L., Li, G., & Li, M. (2026). RFM-UNet: Hybrid Frequency–Mamba UNet for Remote-Sensing Road Extraction. Remote Sensing, 18(15), 2546. https://doi.org/10.3390/rs18152546

Vancouver

Song P, Yu P, Zhong X, Wu S, Tong J, He Y, et al. RFM-UNet: Hybrid Frequency–Mamba UNet for Remote-Sensing Road Extraction. Remote Sensing. 2026;18(15):2546. doi:10.3390/rs18152546

BibTeX
@article{Song2026,
  author = {Song, Pu and Yu, Peng and Zhong, Xiaojing and Wu, Shuizhen and Tong, Junbin and He, Yuanrong and Zhang, Lujun and Li, Guangchun and Li, Mengmeng},
  title = {RFM-UNet: Hybrid Frequency–Mamba UNet for Remote-Sensing Road Extraction},
  journal = {Remote Sensing},
  year = {2026},
  volume = {18},
  number = {15},
  pages = {2546},
  doi = {10.3390/rs18152546}
}
Kostenlos starten →