A Self-Supervised Method for Attenuating Seismic Random and Tracewise Coherent Noise Under the Nonpixelwise Independence Assumption
- 20 May 2025
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 63 (01962892) , 1-12
Abstract
The attenuation of seismic field noise using self-supervised deep learning (DL) has gained attention due to its label-free training process. However, common self-supervised methods are limited by the pixelwise independence assumption, which does not align with field seismic noise characteristics, and suffer from signal leakage due to receptive fields containing inherent blind spots or traces. In this article, we propose a self-supervised random noise attenuation method based on the nonpixelwise independence assumption. By considering the spatial correlation map of field noise, we extend the blind spot to a generalized blind neighborhood, ensuring that the prediction pixel is not influenced by neighboring pixels with noise correlation greater than zero. The blind neighborhood size controls how much spatial correlation is disrupted, allowing our method to handle random noise with varying spatial correlation. Since larger blind neighborhoods may lead to signal loss, we introduce an automatic tradeoff between noise correlation disruption and signal preservation during training. Experiments on real seismic noise attenuation (including random and tracewise coherent noise) demonstrate the superiority of our method in destroying the spatial coherence of noise and preventing useful signal leakage. Our code is available at https://github.com/mengchuangji/SSDnKeywords
Funding Information
- National Key Research and Development Program of China (2020YFA0713403, 2020YFA0713400)
- China National Post-Doctoral Program for Innovative Talents (BX20230284)
- China Post-Doctoral Science Foundation (2024M762608)
- Post-Doctoral Research Project of Shaanxi Province
- Fundamental Research Funds for the Central Universities (xzy012024089)
- Shaanxi Provincial Natural Science Basic Research Program Youth Project (2025JC-YBQN-335)
This publication has 37 references indexed in Scilit:
- Spatially Adaptive Self-Supervised Learning for Real-World Image Denoising2023
- The potential of self-supervised networks for random noise suppression in seismic dataArtificial Intelligence in Geosciences, 2021
- Construction of Optimal Basic Wavelet via AIDNN and Its Application in Seismic Data AnalysisIEEE Geoscience and Remote Sensing Letters, 2020
- Noise2Void - Learning Denoising From Single Noisy Images2019
- Image Super-Resolution Using Very Deep Residual Channel Attention Networks2018
- Deep Image Prior2018
- Application of Principal Component Analysis in Weighted Stacking of Seismic DataIEEE Geoscience and Remote Sensing Letters, 2017
- U-Net: Convolutional Networks for Biomedical Image Segmentation2015
- Seismic Data Analysis Techniques in Hydrocarbon Exploration2014
- Robust Matrix Factorization with Unknown Noise2013