A seismic random noise suppression method based on CNN-Mamba

Authors

WEI Xiujuan, School of Geophysics, Chengdu University of Technology, Chengdu 610059, ChinaFollow
LIU Xingye, School of Geophysics, Chengdu University of Technology, Chengdu 610059, China
ZHOU Huailai, School of Geophysics, Chengdu University of Technology, Chengdu 610059, ChinaFollow

Abstract

Background Seismic random noise suppression is recognized as a key step to improve the quality of seismic data. Data-driven deep learning provides an intelligent solution for the noise suppression. However, mainstream random noise intelligent methods based on convolutional neural networks (CNNs) are constrained by their local receptive fields. This limitation results in insufficient collaborative optimization between local details and macroscopic structures during denoising, further reducing the noise suppression accuracy. Transformer models, which are widely applied to global feature extraction, can effectively capture long-distance dependencies through the self-attention mechanism, theoretically overcoming the limitations of CNNs in global modeling. However, these models face challenges such as slow computation, high resource consumption, and limited applications.Objective and Methods To address these issues, this study proposed a CMUNet seismic random noise suppression network that integrates CNN and Mamba. Based on the 2D-selective-scan (SS2D) mechanism, which can traverse the input data along horizontal and vertical directions, a global dynamic system was constructed using state-space equations. This enabled the trans-scale feature extraction of the spatiotemporal characteristics of seismic data. The hardware-aware parallel algorithm of Mamba was employed to reduce the computational resource consumption, thus ensuring the denoising performance while enhancing computational efficiency. Targeting the characteristics of seismic data, this study designed a CNN-Mamba hybrid block to construct hierarchical feature extraction pathways in the UNet encoder. Specifically, the CNN in a shallow layer focused on local noise pattern recognition, while Mamba in a deep layer was used to capture the correlations of large-scale geological structures. Additionally, residual channel attention gating was further introduced to enhance the feature separability between effective signals and noise. Results and Conclusions The results indicate that for synthetic data, the proposed CMUNet network increased the signal-to-noise ratio (RS/N), peak signal-to-noise ratio (RPSN), and structural similarity by 2.4 dB, 2.4 dB, and 0.005 6, respectively compared to UNet. These results suggest that the CMUNet network enables effective random noise suppression and preserves effective signals. This network was applied to field seismic data. An image-based local similarity analysis reveals that the network yielded low local similarity, further corroborating that it causes minimal damage to effective signals and exhibits superior amplitude preservation. Therefore, the CMUNet network proposed in this study holds great potential for application.

Keywords

seismic random noise suppression, deep learning, convolutional neural network (CNN), state-space model (SSM), Mamba

DOI

10.12363/issn.1001-1986.24.09.0601

Recommended Citation

WEI Xiujuan, LIU Xingye, ZHOU Huailai, et al. (2025) "A seismic random noise suppression method based on CNN-Mamba," Coal Geology & Exploration: Vol. 53: Iss. 5, Article 18.
DOI: 10.12363/issn.1001-1986.24.09.0601
Available at: https://cge.researchcommons.org/journal/vol53/iss5/18

Reference

[1] 国胧予，刘财，刘洋，等. 基于f–x域流式预测滤波器的地震随机噪声衰减方法[J]. 地球物理学报，2020，63(1)：329−338.

GUO Longyu，LIU Cai，LIU Yang，et al. Seismic random noise attenuation based on streaming prediction filter in the f–x domain[J]. Chinese Journal of Geophysics，2020，63(1)：329−338.

[2] 石战战，庞溯，王元君，等. 基于f–x域TV正则化的共偏移距道集随机噪声压制方法[J]. 地球物理学进展，2022，37(3)：1148−1158.

SHI Zhanzhan，PANG Su，WANG Yuanjun，et al. Random noise attenuation of common offset gathers by f–x TV regularization[J]. Progress in Geophysics，2022，37(3)：1148−1158.

[3] WANG Zhiyong，LIU Guochang，LI Chao，et al. Random noise attenuation of 3D multicomponent seismic data using a fast adaptive prediction filter[J]. Geophysics，2024，89(3)：V263−V280.

[4] ABMA R，CLAERBOUT J. Lateral prediction for noise attenuation by t–x and f–x techniques[J]. Geophysics，1995，60(6)：1887−1896.

[5] 俞岱，黄德智，孙渊，等. 基于多t–x域联合的多震源地震数据混叠噪声分离方法[J]. 煤田地质与勘探，2021，49(5)：209−219.

YU Dai，HUANG Dezhi，SUN Yuan，et al. Blending interference noise separation method of simultaneous source seismic data based on multi–t–x domain combination[J]. Coal Geology & Exploration，2021，49(5)：209−219.

[6] 刘强. 随采地震噪声衰减研究[J]. 煤田地质与勘探，2019，47(3)：25−28.

LIU Qiang. Study on noise attenuation of seismic while mining[J]. Coal Geology & Exploration，2019，47(3)：25−28.

[7] 武国宁，于萌萌，王君仙，等. 应用平稳小波变换与深度残差网络压制地震随机噪声[J]. 石油地球物理勘探，2022，57(1)：43−51.

WU Guoning，YU Mengmeng，WANG Junxian，et al. Seismic random noise attenuation based on stationary wavelet transform and deep residual neural network[J]. Oil Geophysical Prospecting，2022，57(1)：43−51.

[8] 周东红，周建科，夏同星，等. 三参数小波变换自适应阈值压制地震数据高频随机噪声[J]. 地球物理学报，2023，66(5)：2095−2111.

ZHOU Donghong，ZHOU Jianke，XIA Tongxing，et al. Suppression of high frequency random noise in seismic data by self–adaptive threshold of three parameter wavelet transform[J]. Chinese Journal of Geophysics，2023，66(5)：2095−2111.

[9] LI Jinghe，FENG Naixing，RAN Mengkun. Processing step–based adaptive seismic denoising method of transformation domain hybrid technique[J]. IEEE Geoscience and Remote Sensing Letters，2021，19：7502505.

[10] 路鹏飞，郭爱华，何月顺，等. 曲波变换与傅里叶变换联合压制面波方法研究[J]. 地球物理学进展，2020，35(6)：2181−2187.

LU Pengfei，GUO Aihua，HE Yueshun，et al. Research on surface wave suppression combined with Curvelet transform and Fourier transform[J]. Progress in Geophysics，2020，35(6)：2181−2187.

[11] 唐欢欢，毛伟建. 基于多路径Radon变换的地震数据噪声压制和波型分离方法研究[J]. 地球物理学报，2022，65(1)：333−348.

TANG Huanhuan，MAO Weijian. Multi–path Radon transform and its application in denoising and wave field separation[J]. Chinese Journal of Geophysics，2022，65(1)：333−348.

[12] 董烈乾，周恒，郭善力，等. 一种改进型seislet域迭代阈值压制混叠噪声方法[J]. 物探与化探，2020，44(3)：568−572.

DONG Lieqian，ZHOU Heng，GUO Shanli，et al. An optimized blending noise suppression based on seislet domain iterative threshold denoising approach[J]. Geophysical and Geochemical Exploration，2020，44(3)：568−572.

[13] LIN Rongzhi，BAHIA B，SACCHI M D. Iterative deblending of simultaneous–source seismic data via a robust singular spectrum analysis filter[J]. IEEE Transactions on Geoscience and Remote Sensing，2021，60：5904110.

[14] FENG Qiankun，LI Yue. Denoising deep learning network based on singular spectrum analysis：DAS seismic data denoising with multichannel SVDDCNN[J]. IEEE Transactions on Geoscience and Remote Sensing，2021，60：5902911.

[15] WU Juan，CHEN Qingli，GUI Zhixian，et al. Fast dictionary learning for 3D simultaneous seismic data reconstruction and denoising[J]. Journal of Applied Geophysics，2021，194：104446.

[16] 毛世榕，史水平，玉壮基，等. 基于自适应噪声完全集合经验模态分解算法和Hurst指数的地震数据去噪方法[J]. 地震学报，2023，45(2)：258−270.

MAO Shirong，SHI Shuiping，YU Zhuangji，et al. A seismic data denoising method based on complete ensemble empirical mode decomposition with adaptive noise and Hurst exponent[J]. Acta Seismologica Sinica，2023，45(2)：258−270.

[17] 曾爱平，张嘉玮，任恩明，等. 基于VMD和SVM的煤厚预测方法研究[J]. 煤田地质与勘探，2021，49(6)：243−250.

ZENG Aiping，ZHANG Jiawei，REN Enming，et al. Research on the coal thickness prediction method based on VMD and SVM[J]. Coal Geology & Exploration，2021，49(6)：243−250.

[18] GAO Lei，LIANG Dongsheng，MIN Fan. Unsupervised denoising for seismic data with complementary mask blind spot strategy[J]. Journal of Applied Geophysics，2024，221：105307.

[19] GAO Yang，ZHAO Pingqi，LI Guofa，et al. Seismic noise attenuation by signal reconstruction：An unsupervised machine learning approach[J]. Geophysical Prospecting，2021，69(5)：984−1002.

[20] ZHANG Kai，ZUO Wangmeng，CHEN Yunjin，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.

[21] RONNEBERGER O，FISCHER P，BROX T. U–Net：Convolutional networks for biomedical image segmentation[M]//Medical Image Computing and Computer–Assisted Intervention–MICCAI 2015. Cham：Springer International Publishing，2015：234–241.

[22] DONG Xintong，LI Yue，ZHONG Tie，et al. Random and coherent noise suppression in DAS–VSP data by using a supervised deep learning method[J]. IEEE Geoscience and Remote Sensing Letters，2020，19：8001605.

[23] 唐杰，韩盛元，刘英昌，等. 基于去噪卷积神经网络的面波噪声压制方法[J]. 石油物探，2022，61(2)：245−252.

TANG Jie，HAN Shengyuan，LIU Yingchang，et al. Seismic surface wave attenuation based on denoising convolutional neural networks[J]. Geophysical Prospecting for Petroleum，2022，61(2)：245−252.

[24] KAUR H，FOMEL S，PHAM N. Seismic ground–roll noise attenuation using deep learning[J]. Geophysical Prospecting，2020，68(7)：2064−2077.

[25] LI Yue，WANG Yuying，WU Ning. Noise suppression method based on multi–scale Dilated Convolution Network in desert seismic data[J]. Computers & Geosciences，2021，156：104910.

[26] 杨翠倩，周亚同，何昊，等. 基于全局上下文和注意力机制深度卷积神经网络的地震数据去噪[J]. 石油物探，2021，60(5)：751−762.

YANG Cuiqian，ZHOU Yatong，HE Hao，et al. Global context and attention–based deep convolutional neural network for seismic data denoising[J]. Geophysical Prospecting for Petroleum，2021，60(5)：751−762.

[27] 王钰清，陆文凯，刘金林，等. 基于数据增广和CNN的地震随机噪声压制[J]. 地球物理学报，2019，62(1)：421−433.

WANG Yuqing，LU Wenkai，LIU Jinlin，et al. Random seismic noise attenuation based on data augmentation and CNN[J]. Chinese Journal of Geophysics，2019，62(1)：421−433.

[28] 高好天，孙宁娜，孙可奕，等. DnCNN和U–Net对地震随机噪声压制的对比分析[J]. 地球物理学进展，2021，36(6)：2441−2453.

GAO Haotian，SUN Ningna，SUN Keyi，et al. Comparative analysis of DnCNN and U–Net on suppression of seismic random noise[J]. Progress in Geophysics，2021，36(6)：2441−2453.

[29] 高磊，沈侯森，闵帆. 基于密集扩张卷积残差网络的地震数据随机噪声压制方法[J]. 石油物探，2023，62(4)：655−668.

GAO Lei，SHEN Housen，MIN Fan. Random noise suppression method based on dense dilated convolutional residual networks in seismic data[J]. Geophysical Prospecting for Petroleum，2023，62(4)：655−668.

[30] ZHONG Tie，CHENG Ming，DONG Xintong，et al. Seismic random noise suppression by using deep residual U–Net[J]. Journal of Petroleum Science and Engineering，2022，209：109901.

[31] WANG Hongzhou，LIN Jun，LI Yue，et al. Self–supervised pretraining transformer for seismic data denoising[J]. IEEE Transactions on Geoscience and Remote Sensing，2024，62：5907525.

[32] GAO Lei，SHEN Housen，MIN Fan. Swin Transformer for simultaneous denoising and interpolation of seismic data[J]. Computers & Geosciences，2024，183：105510.

[33] DING Mu，ZHOU Yatong，CHI Yue. Seismic signal denoising using Swin–Conv–UNet[J]. Journal of Applied Geophysics，2024，223：105355.

[34] GRAY S H，MARFURT K J. Migration from topography：Improving the near–surface image[J]. Canadian Journal of Exploration Geophysics，1995，31(1/2)：18−24.

[35] ETGEN J，REGONE C. Strike shooting，dip shooting，widepatch shooting：Does prestack depth migration care?A model study[C]//SEG Technical Program Expanded Abstracts 1998. Society of Exploration Geophysicists，1998：2092.

[36] GAJERA B，KAPIL S R，ZIAEI D，et al. CT–scan denoising using a Charbonnier loss generative adversarial network[J]. IEEE Access，2021，9：84093−84109.