EUVIMEDEUVIMED entdecken
Crossref

AFF-Net: Adaptive Frequency-Domain Fusion in a Dual-Stream Transformer With Wavelet-Assisted Alignment for Remote Sensing Image Segmentation

Qingyang Ren, Tiancheng Ma, Leiguang Wang, Chen Zheng

IEEE Geoscience and Remote Sensing Letters · 2026 · 23 · 6011305-6011305

DOI: 10.1109/lgrs.2026.3695313 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Geoscience and Remote Sensing Letters, 2026. Verfasst von Qingyang Ren und 3 weiteren Autor:innen. Die Arbeit nennt 20 Literaturangaben. Gefördert von: Program for Innovative Research Team (in Science and Technology) in University of Henan Province, State Key Laboratory of Spatial Datum.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Remote sensing semantic segmentation is essential for geoscientific analysis. Recently, although multimodal image fusion has achieved notable improvements over single-image approaches, existing methods still suffer from insufficient utilization of the observed information, especially in the alignment and fusion of multimodal detailed information. To address the above issues, a novel adaptive frequency-domain fusion network (AFF-Net) is proposed in this letter to achieve effective fusion of visible images (VISs) and digital surface model (DSM) data through a frequency-domain fusion (FDF) module, and to enhance the recognition capability of target boundaries via a semantic-guided detail alignment (SDA) strategy. In particular, the FDF module conducts an adaptive frequency-wise alignment and fusion scheme between VISs and DSM by decomposing them into low- and high-frequency subbands, enabling the model to capture more fine-grained complementary information across various frequency domains, rather than coarse-grained fusion in the spatial domain. The SDA strategy further restores high-frequency boundary details while suppressing noise to enable precise object delineation. AFF-Net was evaluated on the ISPRS Vaihingen and Potsdam datasets, achieving mIoU scores of 84.87% and 87.11%, as well as mF1 scores of 91.61% and 92.93%, respectively. These results show better performance over state-of-the-art (SOTA) methods and verify the effectiveness and superiority of the proposed method. The code is available at https://github.com/lanmao-create/AFF-Net

Förderung

  • Program for Innovative Research Team (in Science and Technology) in University of Henan Province
  • State Key Laboratory of Spatial Datum

Zitieren

APA

Ren, Q., Ma, T., Wang, L., & Zheng, C. (2026). AFF-Net: Adaptive Frequency-Domain Fusion in a Dual-Stream Transformer With Wavelet-Assisted Alignment for Remote Sensing Image Segmentation. IEEE Geoscience and Remote Sensing Letters, 23, 6011305-6011305. https://doi.org/10.1109/lgrs.2026.3695313

Vancouver

Ren Q, Ma T, Wang L, Zheng C. AFF-Net: Adaptive Frequency-Domain Fusion in a Dual-Stream Transformer With Wavelet-Assisted Alignment for Remote Sensing Image Segmentation. IEEE Geoscience and Remote Sensing Letters. 2026;23:6011305-6011305. doi:10.1109/lgrs.2026.3695313

BibTeX
@article{Ren2026,
  author = {Ren, Qingyang and Ma, Tiancheng and Wang, Leiguang and Zheng, Chen},
  title = {AFF-Net: Adaptive Frequency-Domain Fusion in a Dual-Stream Transformer With Wavelet-Assisted Alignment for Remote Sensing Image Segmentation},
  journal = {IEEE Geoscience and Remote Sensing Letters},
  year = {2026},
  volume = {23},
  pages = {6011305--6011305},
  doi = {10.1109/lgrs.2026.3695313}
}
Kostenlos starten →