EUVIMEDEUVIMED entdecken
Crossref

RGCDNet: Radiometric–Geometric Collaborative Learning for Change Detection in Remote Sensing

Xia Zhang, Jia Liu, Wenhua Zhang, Fang Liu, Jingxiang Yang, Yong Li, Liang Xiao

IEEE Transactions on Geoscience and Remote Sensing · 2026 · 64 · 1-17

DOI: 10.1109/tgrs.2026.3683370 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Geoscience and Remote Sensing, 2026. Verfasst von Xia Zhang und 6 weiteren Autor:innen. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 65 Literaturangaben. Gefördert von: National Nature Science Foundation of China, Frontier Technologies Research and Development Program of Jiangsu Province, Fundamental Research Funds for the Central Universities.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

In remote sensing change detection (CD), dealing with the interference from pseudochanges such as those caused by illumination variations and geometric registration errors is a key challenge. Integrating physical characteristics of radiation and geometry into a deep network can help to reduce such pseudochanges. As a consequence, this article proposes a novel radiometric–geometric collaborative learning network for CD in remote sensing images (RGCDNet), which incorporates a multisource radiometric shadow suppression (MSR-SS) module and a collaborative registration module (CRM) in order to effectively mitigate the pseudochanges caused by shadows and geometric registration errors. MSR-SS uses the radiometric characteristics of vegetation and water bodies to enhance the separability between shadows and other land features, and constructs spectral separation and suppression processes with a trainable network to minimize false positives (FPs). CRM, meanwhile, achieves collaborative optimization of geometric registration and CD by applying bidirectional deformation to feature maps of various stages and integrating a unified flow field through weighted fusion. Experimental results show that, compared with existing state-of-the-art methods, our RGCDNet significantly improves detection accuracy and robustness. Specifically, we obtained $F1$ -scores of 91.31%, 83.32%, and 94.21% on the LEVIR-CD, SYSU-CD, and WHU-CD datasets, respectively. The code of this work can be found on https://github.com/4043-lab/Change-detection

Förderung

  • National Nature Science Foundation of China
  • Frontier Technologies Research and Development Program of Jiangsu Province
  • Fundamental Research Funds for the Central Universities
  • Open Research Fund of Jiangsu Key Laboratory of Visual Sensing and Intelligent Perception
  • Natural Science Foundation of Jiangsu Higher Education Institutions of China

Zitieren

APA

Zhang, X., Liu, J., Zhang, W., Liu, F., Yang, J., Li, Y., & Xiao, L. (2026). RGCDNet: Radiometric–Geometric Collaborative Learning for Change Detection in Remote Sensing. IEEE Transactions on Geoscience and Remote Sensing, 64, 1-17. https://doi.org/10.1109/tgrs.2026.3683370

Vancouver

Zhang X, Liu J, Zhang W, Liu F, Yang J, Li Y, et al. RGCDNet: Radiometric–Geometric Collaborative Learning for Change Detection in Remote Sensing. IEEE Transactions on Geoscience and Remote Sensing. 2026;64:1-17. doi:10.1109/tgrs.2026.3683370

BibTeX
@article{Zhang2026,
  author = {Zhang, Xia and Liu, Jia and Zhang, Wenhua and Liu, Fang and Yang, Jingxiang and Li, Yong and Xiao, Liang},
  title = {RGCDNet: Radiometric–Geometric Collaborative Learning for Change Detection in Remote Sensing},
  journal = {IEEE Transactions on Geoscience and Remote Sensing},
  year = {2026},
  volume = {64},
  pages = {1--17},
  doi = {10.1109/tgrs.2026.3683370}
}
Kostenlos starten →