EUVIMEDEUVIMED entdecken
Crossref

Robust Seismic Denoising Framework: A GAN-Based Approach With Multiscale Feature Fusion and Signal Preservation

Kun Li, Jin Yuan, Wei Ge, Xingyao Yin, Zhaoyun Zong

IEEE Geoscience and Remote Sensing Letters · 2026 · 23 · 1-5

DOI: 10.1109/lgrs.2026.3676613 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Geoscience and Remote Sensing Letters, 2026. Verfasst von Kun Li und 4 weiteren Autor:innen. Die Arbeit nennt 15 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Fund of the State Key Laboratory of Deep Oil and Gas, China University of Petroleum, Fundamental Research Funds for the Central Universities.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Seismic data acquisition is often affected by complex field conditions and equipment precision limitations, leading to the generation of multiple types of noise that reduce the accuracy of reservoir inversion and geological interpretation. Deep learning techniques have begun to be applied in seismic data denoising due to their efficient feature learning capabilities. However, the complex characteristics of seismic noise pose challenges for conventional convolutional neural networks in learning diverse noise features and achieving effective feature fusion. We propose a novel seismic data denoising methodology based on generative adversarial networks (SD-GAN). The established denoising framework achieves noise distribution prediction through adversarial learning while integrating a deep residual convolutional network (DnCNN) with attention mechanisms to precisely learn diverse noise distributions and enable feature fusion. The DnCNN incorporates deformable convolutions that dynamically adjust convolution kernel shapes to adapt to different seismic signal characteristics, enhancing the network’s modeling capability for complex geological structures. To ensure effective frequency-domain denoising performance, an adaptive frequency-domain loss term is introduced in the loss function to maximize the preservation of seismic signal characteristics. Validation through synthetic seismic data experiments and real seismic data applications demonstrates that this method exhibits excellent robustness and feasibility across various denoising scenarios.

Förderung

  • National Natural Science Foundation of China
  • Fund of the State Key Laboratory of Deep Oil and Gas, China University of Petroleum
  • Fundamental Research Funds for the Central Universities
  • China National Petroleum Corporation (CNPC) Innovation Fund

Zitieren

APA

Li, K., Yuan, J., Ge, W., Yin, X., & Zong, Z. (2026). Robust Seismic Denoising Framework: A GAN-Based Approach With Multiscale Feature Fusion and Signal Preservation. IEEE Geoscience and Remote Sensing Letters, 23, 1-5. https://doi.org/10.1109/lgrs.2026.3676613

Vancouver

Li K, Yuan J, Ge W, Yin X, Zong Z. Robust Seismic Denoising Framework: A GAN-Based Approach With Multiscale Feature Fusion and Signal Preservation. IEEE Geoscience and Remote Sensing Letters. 2026;23:1-5. doi:10.1109/lgrs.2026.3676613

BibTeX
@article{Li2026,
  author = {Li, Kun and Yuan, Jin and Ge, Wei and Yin, Xingyao and Zong, Zhaoyun},
  title = {Robust Seismic Denoising Framework: A GAN-Based Approach With Multiscale Feature Fusion and Signal Preservation},
  journal = {IEEE Geoscience and Remote Sensing Letters},
  year = {2026},
  volume = {23},
  pages = {1--5},
  doi = {10.1109/lgrs.2026.3676613}
}
Kostenlos starten →