Semi-Supervised DAS VSP Data Denoising Using Signal and Noise Distribution Difference
Top Cited Papers
- 4 December 2024
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 62 (01962892) , 1-14
Abstract
Distributed acoustic sensing (DAS), an emerging technology for signal acquisition, has been progressively applied to collect vertical seismic profile (VSP) data. Unfortunately, the obtained DAS VSP data are usually contaminated by various complex noise, which poses a major obstacle to subsequent processing. Therefore, suppressing the noise in the DAS VSP data is a critical step. With the development of neural network, deep learning is widely used for seismic data denoising. However, supervised learning-based denoising methods require massive amounts of training dataset with labels. The lack of labeled dataset limits the performance of supervised learning methods. The recently proposed unsupervised learning-based denoising methods reduce the reliance on labeled data, but they are not suitable for processing seismic data containing multiple types of complex noise. In this study, we propose a semi-supervised denoising network that contains both supervised and unsupervised paths. Supervised path is trained using synthetic dataset to extract rich signal feature. Unsupervised path exploits the distribution difference between signal and noise, using field dataset to extract realistic and accurate signal feature. The backbone network consists of a three-layer pyramid structure and incorporates multi-scale fusion strategy to improve network performance. The idea of semi-supervised learning reduces the reliance on labeled data, takes full advantage of the distribution characteristic of the field data, and benefits the generalization ability of the denoising model. Experimental results on one synthetic data and four field DAS VSP data demonstrate that the proposed method obtains competitive performance in intense noise suppression and effective signal recovery.Keywords
Funding Information
- National Natural Science Foundation of China (42174153, 42230805)
- Postdoctoral Innovation Talents Support Program (BX20230141)
- Graduate Innovation Fund of Jilin University (2024CX093)
This publication has 38 references indexed in Scilit:
- Noisy2Noisy: Denoise Pre-Stack Seismic Data Without Paired Training Data With LabelsIEEE Geoscience and Remote Sensing Letters, 2022
- Multi-Stage Progressive Image Restoration2021
- Syn2Real Transfer Learning for Image Deraining Using Gaussian Processes2020
- Random Noise Attenuation Based on Residual Convolutional Neural Network in Seismic DatasetsIEEE Access, 2020
- Unsupervised Seismic Random Noise Attenuation Based on Deep Convolutional Neural NetworkIEEE Access, 2019
- Low-Frequency Noise Suppression of Desert Seismic Data Based on Variational Mode Decomposition and Low-Rank Component ExtractionIEEE Geoscience and Remote Sensing Letters, 2019
- Semi-Supervised Transfer Learning for Image Rain Removal2019
- Empirical Low-Rank Approximation for Seismic Noise AttenuationIEEE Transactions on Geoscience and Remote Sensing, 2017
- Curvature-Varying Hyperbolic Trace TFPF for Seismic Random Noise AttenuationIEEE Geoscience and Remote Sensing Letters, 2015
- Deep learningNature, 2015