Self-supervised seismic random noise attenuation based on conditional adversarial networks
-
摘要:
针对地震数据标注困难，提出基于改进的条件对抗网络的自监督随机噪声压制方法.训练过程分为2步：（1）向合成地震记录混入随机噪声构造含噪声-纯净训练集，采用监督学习策略，通过改进的条件生成对抗网络学习地震数据的有效特征；（2）借助自监督损失函数，利用目标域实际数据对预训练模型进行微调.2步训练法利用了源域合成地震记录与目标域实际地震数据之间的相似性，将源域学习到的模型迁移到目标域，实现地震数据自适应盲去噪.理论模型和实际地震数据试算结果验证所提方法具有较好的应用效果.
Abstract:Aiming at the difficulty of labeling seismic data, an improved self-supervised random noise suppression method based on conditional adversarial network is proposed. The training process is divided into two steps: (1) Random noise is mixed into the synthetic seismograms to construct paired noise-pure training sets. Supervised learning strategy is adopted to learn good representations of the seismic data through improved conditional adversarial network. (2) With the help of the self-supervised loss function, the practical data of the target domain is used to fine-tune the pre-training model. The 2-step training method takes advantage of the similarity between the synthetic seismograms in the source domain and the practical seismic data in the target domain, transfers the model learned from the source domain to the target domain, and realizes the adaptive blind denoising of seismic data. The experimental results of the synthetic seismograms and practical seismic data verify that the proposed method has good performance.
-
-
图 2
加入自监督机制后的生成器模型
Figure 2.
Illustration of the generator of the self-supervised CGAN
图 3
不同噪声强度下3种模型相对误差收敛结果
Figure 3.
Relative error convergence results of three models under different noise intensity
图 4
生成器(a)和鉴别器(b)损失对比
Figure 4.
Comparison of the loss functions of the generator (a) and the discriminator (b)
图 6
2种模型的去噪结果对比分析
Figure 6.
Comparative analysis of denoising results of the two models
图 7
预训练模型与微调模型去噪结果对比分析
Figure 7.
Comparative analysis of denoising results of pre-training model and fine-tuning model
图 8
单炮对比预训练模型与微调模型去噪结果
Figure 8.
Denoising results of the pre-training and fine-tuning models with the shot gather
图 9
自监督迁移学习去噪性能分析
Figure 9.
Performance analysis of self-supervised transfer Learning denoising
图 10
实际地震资料去噪结果对比分析
Figure 10.
Comparative analysis of denoising results of the practical seismic data
表 1
合成地震记录
Table 1.
Synthetic seismogram
序号 数据名 数据网址 1 Model94 https://software.seg.org/datasets/2D/Model_1994/ 2 Amoco statics test dataset https://wiki.seg.org/wiki/1994_BP_statics_benchmark_model 3 BP 2.5D migration benchmark model https://wiki.seg.org/wiki/1997_BP_2.5D_migration_benchmark_model 4 BP velocity estimation benchmark model https://wiki.seg.org/wiki/2004_BP_velocity_estimation_benchmark_model 5 BP Anisotropic Velocity Benchmark https://wiki.seg.org/wiki/2007_BP_Anisotropic_Velocity_Benchmark 
下载: 导出CSV
-
Abma R, Claerbout J. 1995. Lateral prediction for noise attenuation by t-x and f-x techniques. Geophysics, 60(6): 1887-1896. doi: 10.1190/1.1443920
Batson J, Royer L. 2019. Noise2Self: blind denoising by self-supervision. //Proceedings of the 36th International Conference on Machine Learning. Long Beach: PMLR, 524-533.
Bekara M, van der Baan M. 2009. Random and coherent noise attenuation by empirical mode decomposition. Geophysics, 74(5): V89-V98. doi: 10.1190/1.3157244
Chen D W, Yang W Y, Wei X J, et al. 2021. Research on first-break automatic picking based on an improved U-Net network. Progress in Geophysics (in Chinese), 36(4): 1493-1503, doi: 10.6038/pg2021EE0235.
Chen Y K, Ma J T. 2014. Random noise attenuation by f-x empirical-mode decomposition predictive filtering. Geophysics, 79(3): V81-V91. doi: 10.1190/geo2013-0080.1
Chen Y K, Ma J W, Fomel S. 2016. Double-sparsity dictionary for seismic noise attenuation. Geophysics, 81(2): V103-V116. doi: 10.1190/geo2014-0525.1
Deng L, Yuan S Y, Wang S X. 2017. Sparse Bayesian learning-based seismic denoise by using physical wavelet as basis functions. IEEE Geoscience and Remote Sensing Letters, 14(11): 1993-1997. doi: 10.1109/LGRS.2017.2745564
Ganin Y, Ustinova E, Ajakan H, et al. 2016. Domain-adversarial training of neural networks. The Journal of Machine Learning Research, 17(1): 2096-2030.
Huang T, Li S J, Jia X, et al. 2021. Neighbor2Neighbor: self-supervised denoising from single noisy images. //2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Nashville: IEEE, 14776-14785.
Isola P, Zhu J Y, Zhou T H, et al. 2017. Image-to-image translation with conditional adversarial networks. //2017 IEEE Conference on Computer Vision and Pattern Recognition. Honolulu: IEEE, 5967-5976.
Kingma D P, Ba J. 2015. Adam: a method for stochastic optimization. //3rd International Conference on Learning Representations. San Diego: ICLR.
LeCun Y, Bengio Y, Hinton G. 2015. Deep learning. Nature, 521(7553): 436-444. doi: 10.1038/nature14539
Lehtinen J, Munkberg J, Hasselgren J, et al. 2018. Noise2Noise: learning image restoration without clean data. //Proceedings of the 35th International Conference on Machine Learning. Stockholm: PMLR, 2965-2974.
Liu G C, Chen X H. 2013. Noncausal f-x-y regularized nonstationary prediction filtering for random noise attenuation on 3D seismic data. Journal of Applied Geophysics, 93: 60-66. doi: 10.1016/j.jappgeo.2013.03.007
Liu G C, Chen X H, Li J Y, et al. 2011. Seismic noise attenuation using nonstationary polynomial fitting. Applied Geophysics, 8(1): 18-26. doi: 10.1007/s11770-010-0244-2
Liu Y, Fomel S, Liu C. 2015. Signal and noise separation in prestack seismic data using velocity-dependent seislet transform. Geophysics, 80(6): WD117-WD128. doi: 10.1190/geo2015-0016.1
Loshchilov I, Hutter F. 2019. Decoupled weight decay regularization. //7th International Conference on Learning Representations. New Orleans: ICLR.
Ma H T, Yao H Y, Li Y, et al. 2020. Deep residual encoder-decoder networks for desert seismic noise suppression. IEEE Geoscience and Remote Sensing Letters, 17(3): 529-533. doi: 10.1109/LGRS.2019.2925062
Neelamani R, Baumstein A, Ross W S. 2010. Adaptive subtraction using complex-valued curvelet transforms. Geophysics, 75(4): V51-V60. doi: 10.1190/1.3453425
Qiu C Y, Wu B Y, Liu N H, et al. 2021. Deep learning prior model for unsupervised seismic data random noise attenuation. IEEE Geoscience and Remote Sensing Letters, 19: 7502005, doi: 10.1109/LGRS.2021.3053760.
Reddi S J, Kale S, Kumar S. 2018. On the convergence of Adam and beyond. //6th International Conference on Learning Representations. Vancouver : ICLR.
Saad O M, Chen Y K. 2020. Deep denoising autoencoder for seismic random noise attenuation. Geophysics, 85(4): V367-V376. doi: 10.1190/geo2019-0468.1
Shi Z Z, Zhou H L, Xia Y Q, et al. 2020. Random noise attenuation of common offset gathers using iteratively reweighted ℓ2, 1norm minimization. IEEE Geoscience and Remote Sensing Letters, 17(11): 1988-1992. doi: 10.1109/LGRS.2019.2960294
Si X, Yuan Y J, Si T H, et al. 2019. Attenuation of random noise using denoising convolutional neural networks. Interpretation, 7(3): SE269-SE280. doi: 10.1190/INT-2018-0220.1
Tian Y N, Li Y, Lin H B, et al. 2013. A sparse NMF-SU for seismic random noise attenuation. IEEE Geoscience and Remote Sensing Letters, 10(3): 607-611. doi: 10.1109/LGRS.2012.2215835
Tzeng E, Hoffman J, Zhang N, et al. 2014. Deep domain confusion: maximizing for domain invariance. arXiv: 1412.3474.
Wang Y Q, Lu W K, Liu J L, et al. 2019. Random seismic noise attenuation based on data augmentation and CNN. Chinese Journal of Geophysics (in Chinese), 62(1): 421-433, doi: 10.6038/cjg2019M0385.
Wei R Q, Garcia C, El-Sayed A, et al. 2020. Variations in variational autoencoders-a comparative evaluation. IEEE Access, 8: 153651-153670. doi: 10.1109/ACCESS.2020.3018151
Yang Q, Zhang Y, Dai W Y, et al. 2020. Transfer Learning (in Chinese). Zhuang F Z Trans. Beijing: China Machine Press.
You J C, Xue Y J, Cao J X, et al. 2020. Attenuation of seismic swell noise using convolutional neural networks in frequency domain and transfer learning. Interpretation, 8(4): T941-T952. doi: 10.1190/INT-2019-0303.1
Yu S W, Ma J W. 2018. Deep learning for attenuating random and coherence noise simultaneously. //88th Ann. Internat Mtg., Soc. Expi. Geophys. . Expanded Abstracts.
Yu S W, Ma J W, Wang W L. 2019. Deep learning for denoising. Geophysics, 84(6): V333-V350. doi: 10.1190/geo2018-0668.1
Zeiler M D, Fergus R. 2014. Visualizing and understanding convolutional networks. //13th European Conference on Computer Vision. Zurich: Springer, 818-833.
Zhang K, Zuo W M, Chen Y J, et al. 2017. Beyond a Gaussian denoiser: residual learning of deep CNN for image denoising. IEEE Transactions on Image Processing, 26(7): 3142-3155. doi: 10.1109/TIP.2017.2662206
Zhang Y S, Lin H B, Li Y, et al. 2019. A patch based denoising method using deep convolutional neural network for seismic image. IEEE Access, 7: 156883-156894. doi: 10.1109/ACCESS.2019.2949774
陈德武, 杨午阳, 魏新建, 等. 2021. 一种基于改进的U-Net网络的初至自动拾取研究. 地球物理学进展, 36(4): 1493-1503, doi: 10.6038/pg2021EE0235. http://www.progeophys.cn/article/doi/10.6038/pg2021EE0235
王钰清, 陆文凯, 刘金林, 等. 2019. 基于数据增广和CNN的地震随机噪声压制. 地球物理学报, 62(1): 421 - 433, doi: 10.6038/cjg2019M0385.
杨强, 张宇, 戴文渊, 等. 2020. 迁移学习. 庄福振译. 北京: 机械工业出版社.
-
图(10)
表(1)
计量
- 文章访问数: 1012
- PDF下载数: 66
- 施引文献: 0