|
|
|
| A method for strong noise suppression based on DC-UNet |
ZHOU Hui( ), SUN Cheng-Yu(), LIU Ying-Chang, CAI Rui-Qian |
| School of Geosciences,China University of Petroleum(East China),Qingdao 266580,China |
|
|
|
|
Abstract Seismic data acquired from mature industrial areas frequently contain a large amount of local strong noise with high amplitude due to the continuous operation of production equipment.However,such local strong noise can be hardly suppressed using conventional denoising methods.This study integrated dilated convolution(DC) and U-Net into a DC-UNet network for suppressing local strong noise.For the circular DC blocks at the front end of the DC-Unet network,a circularly expanded DC kernel was used to extract the features of strong noise at different scales,with the receptive field being expanded.Meanwhile,an encoder was used at the back end of the network to extract the features of strong noise and restore the details of strong noise.Subsequently,the DC-UNet network was employed to perform a nonlinear mapping from noisy data to noise.On this basis,strong noise was suppressed by subtracting the learned strong noise from the noisy data.As indicated by the experimental results of synthetic and real data obtained from the training using the PyTorch framework in the GPU environment,the DC-UNet network can effectively suppress the local strong noise and improve the signal-to-noise ratio compared with DnCNN,U-Net,and PCA-UNet networks.
|
|
Received: 13 October 2022
Published: 27 October 2023
|
|
|
|
Corresponding Authors:
SUN Cheng-Yu
E-mail: 2577156309@qq.com;suncy@upc.edu.cn
|
|
|
|
|
Network structure of conventional U-Net
|
|
Circular dilated convolution block
|
|
Network structure of DC-UNet
|
|
Schematic diagram of downsampling
|
|
Actual line seismic data
a—stacking energy on time;b—3 shot seismic data from one of the lines;c—stacking energy on trace
|
|
Layered medium model
|
|
Synthetic seismic record corresponding to the model
|
|
Strong noise in different positions
|
|
Noise-free data(a) and noisy data(b)
|
|
Denoising comparison of different denoising methods after the 30 training
a—DnCNN denoising results;b—U-Net denoising results;c—PCA-UNet denoising results;d—DC-UNet denoising results;e—noise removed by DnCNN;f—noise removed by U-Net;g—noise removed by PCA-UNet;h—noise removed by DC-UNet
|
|
Denoising comparison of different denoising methods after 50 training
a—DnCNN denoising results;b—U-Net denoising results;c—PCA-UNet denoising results;d—DC-UNet denoising results;e—noise removed by DnCNN;f—noise removed by U-Net;g—noise removed by PCA-UNet;h—noise removed by DC-UNet
|
|
Denoising evaluation index
|
|
Compare the actual seismic data denoising results after the 30 training and the corresponding F-K spectrum
a—actual noisy seismic data;b—U-Net denoising results;c—DC-UNet denoising results;d—F-K spectra of noisy data;e—F-K spectra of U-Net denoising results;f—F-K spectra of DC-UNet denoising results
|
|
Compare the actual seismic data denoising results after the 50 training with the corresponding F-K spectrum and noise removal
a—U-Net denoising results;b—DC-UNet denoising results;c—F-K spectrum of U-Net denoising results;d—F-K spectrum of DC-UNet denoising results;e—noise removed by U-Net;f—noise removed by DC-UNet
|
| [1] |
Shahdoosti H R, Rahemi Z. Edge-preserving image denoising using a deep convolutional neural network[J]. Signal Process, 2019, 159(3):20-32.
|
| [2] |
冯兴强, 杨长春, 龙志祎. 基于奇异值分解的f-x-y域滤波方法[J]. 物探与化探, 2005, 29(2):171-173.
|
| [2] |
Feng X Q, Yang C C, Long Z Y. The filtering method in f-x-y domain based on singular value decomposition[J]. Geophysical and Geochemical Exploration, 2005, 29(2):171-173.
|
| [3] |
刘婷婷, 陈阳康. f-x域经验模式分解与多道奇异谱分析相结合去除随机噪声[J]. 石油物探, 2016, 55(1):67-75.
|
| [3] |
Liu T T, Chen Y K. Random noise attenuation based on EMD and MSSA in f-x domain[J]. Geophysical Prospecting for Petroleum, 2016, 55(1):67-75.
|
| [4] |
Canales L. Random noise reduction[C]// 54th Annual International Meeting,SEG,Expanded Abstract, 1984:525-527.
|
| [5] |
贾春梅, 姜国庆, 刘志成, 等. 频域稀疏双曲Radon变换去噪方法[J]. 物探与化探, 2016, 40(3):527-533.
|
| [5] |
Jia C M, Jiang G Q, Liu Z C, et al. Denising method based on sparse hyperbolic Radon transform in the frequency domain[J]. Geophysical and Geochemical Exploration, 2016, 40(3):527-533.
|
| [6] |
彭才, 常智, 朱仕军. 基于曲波变换的地震数据去噪方法[J]. 石油物探, 2008, 47(5):461-464.
|
| [6] |
Peng C, Chang Z, Zhu S J. Noise elimination method based on curvelet transform[J]. Geophysical Prospecting for Petroleum, 2008, 47(5):461-464.
|
| [7] |
刘法启, 张关泉. 小波变换与F-K算法在滤波中的应用[J]. 石油地球物理勘探, 1996, 31(6):782-791.
|
| [7] |
Liu F Q, Zhang G Q. Application of wavelet transform and F-K algorithm in filtering[J]. Oil Geophysical Prospecting, 1996, 31(6):782-791.
|
| [8] |
杨立强, 宋海斌, 郝天珧, 等. 基于二维小波变换的随机噪声压制方法研究[J]. 石油物探, 2005, 44(1):4-6.
|
| [8] |
Yang L Q, Song H B, Hao T Y, et al. Method of 2-D wavelet transform in attenuating random noise[J]. Geophysical Prospecting for Petroleum, 2005, 44(1):4-6.
|
| [9] |
Zhang K, Zuo W, Chen Y, et al. Beyond a Gaussian Denoiser:Residual learning of deep CNN for image denoising[J]. IEEE Transactions on Image Processing, 2017, 26(7):3142-3155.
|
| [10] |
韩卫雪, 周亚同, 池越. 基于深度学习卷积神经网络的地震数据随机噪声去除[J]. 石油物探, 2018, 57(6):862-869.
|
| [10] |
Han W X, Zhou Y D, Chi Y. Deep learning convolutional neural networks for random noise attenuation in seismic data[J]. Geophysical Prospecting for Petroleum, 2018, 57(6):862-869.
|
| [11] |
Alwon S. Generative adversarial networks in seismic data processing[C]// SEG Technical Program Expanded Abstracts, 2018.
|
| [12] |
于四伟, 马坚伟. 基于深度学习的地震噪声压制[C]// SEG北京2018国际地球物理会议暨展览, 2018.
|
| [12] |
Yu S W, Ma J W. Seismic noise suppression based on deep learning[C]// SEG Beijing 2018 International Geophysical Conference and Exhibition, 2018.
|
| [13] |
Wang F, Chen S. Residual learning of deep convolutional neural network for seismic random noise attenuation[J]. IEEE Geosciences and Remote Sensing Letters, 2019, 16(8):1314-1318.
|
| [14] |
张攀龙, 李尧, 张田涛, 等. 基于U-Net深度神经网络的地震数据去噪研究[J]. 金属矿山, 2020(1):200-208.
|
| [14] |
Zhang P L, Li Y, Zhang T T, et al. Study on seismic data denoising based on U-Net deep neural network[J]. Metal Mine, 2020(1):200-208.
|
| [15] |
罗仁泽, 李阳阳. 一种基于RUnet卷积神经网络的地震资料随机噪声压制方法[J]. 石油物探, 2020, 59(1):51-59.
|
| [15] |
Luo R Z, Li Y Y. Random scismic noisc attenuation based on RUnet convolutional neural network[J]. Gcophysical Prospecting for Petroleum, 2020, 59(1):51-59
|
| [16] |
Li H, Yang W, Yong X. Deep learning for ground-roll noise attenuation[C]// SEG Technical Program Expanded Abstracts, 2018.
|
| [17] |
Yu F, Koltun V. Multi-scale context aggregation by dilated convolutions[C]// ICLR, 2016.
|
| [18] |
张华博. 基于深度学习的图像分割研究与应用[D]. 成都: 电子科技大学, 2018.
|
| [18] |
Zhang H B. Research and application of image segmentation by deep learning[D]. Chengdu: University of Electronic Science and Technology of China, 2018.
|
| [19] |
薛海洋. 基于深度学习的图像分割算法研究[D]. 南宁: 广西大学, 2020.
|
| [19] |
Xue H Y. Research on image segmentation algorithm based on deep learning[D]. Nanning: Guangxi University, 2020.
|
| [1] |
LIANG Li-Feng, LIU Xiu-Juan, ZHANG Hong-Bing, CHEN Cheng-Hao, CHEN Jin-Hua. A study of the effect of hyperparameters GRU-CNN hybrid deep learning EI inversion[J]. Geophysical and Geochemical Exploration, 2021, 45(1): 133-139. |
| [2] |
Yi-Chen WANG, Lin-Tao LIU, Hou-Ze XU. The identification of gravity anomaly body based on the convolutional neural network[J]. Geophysical and Geochemical Exploration, 2020, 44(2): 394-400. |
|
|
|
|
