High-precision reconstruction of seismic data based on improved U-Net++

Authors

WANG Minling, School of Geosciences, Guilin University of Technology, Guilin 541004, ChinaFollow
ZHOU Fei, School of Geosciences, Guilin University of Technology, Guilin 541004, China
WANG Honghua, School of Geosciences, Guilin University of Technology, Guilin 541004, ChinaFollow
HE Xiang, School of Geosciences, Guilin University of Technology, Guilin 541004, China
HOU Zhiyang, School of Geosciences, Guilin University of Technology, Guilin 541004, China

Abstract

Objective Conventional reconstruction methods are insufficient for the reconstruction of seismic data with missing consecutive traces, producing a negative impact on subsequent processing accuracy. Hence, this study proposed CU-Net++, a deep learning network based on the U-Net++ architecture combined with the convolutional block attention module (CBAM). Methods During the reconstruction of missing data, the independent decoder for each sub-U-Net in the nested U-Net++ architecture enables the utilization of information from different depths. The long and short skip connections can effectively enhance the network's capability to extract multi-scale features from data. The core innovation of CU-Net++ is the introduction of CBAM, which can enhance the capacity to learn about seismic wave details and edge information, into the U-Net++. This helps improve the network's ability to identify and capture complex seismic wave characteristics. Through the reconstruction tests of simulated and measured data, this study presented a comparative analysis of the reconstruction effects for missing seismic data of the CU-Net++, U-Net++, CU-Net, U-Net, and curvelet-domain projection onto convex sets (POCS) methods from the perspective of F-K spectrum, residual profile, single-trace waveform, mean absolute error (MAE), signal-to-noise ratio (SNR), and peak signal-to-noise ratio (PSNR).Results and Conclusions CU-Net++ delivered the optimum overall performance across various assessment metrics, yielding the lowest reconstruction error. Compared to U-Net++, it reduced the MAE by approximately 51% and improved the SNR and PSNR by 5.87 dB each. Notably, CU-Net++ enables high-precision construction of seismic data with a proportion of consecutively missing traces not exceeding 12%.

Keywords

seismic data reconstruction, deep learning, CU-Net++, nested architecture, multi-scale feature fusion, attention mechanism

DOI

10.12363/issn.1001-1986.25.01.0034

Recommended Citation

WANG Minling, ZHOU Fei, WANG Honghua, et al. (2025) "High-precision reconstruction of seismic data based on improved U-Net++," Coal Geology & Exploration: Vol. 53: Iss. 9, Article 15.
DOI: 10.12363/issn.1001-1986.25.01.0034
Available at: https://cge.researchcommons.org/journal/vol53/iss9/15

Reference

[1] GAO Jianjun，CHEN Xiaohong，LI Jingye，et al. Irregular seismic data reconstruction based on exponential threshold model of POCS method[J]. Applied Geophysics，2010，7(3)：229−238.

[2] 贾永娜，吴杰，王国伟，等. 基于Gabor纹理学习的地震数据重建算法[J]. 石油地球物理勘探，2023，58(3)：617−625.

JIA Yongna，WU Jie，WANG Guowei，et al. A texture feature learning method based on Gabor transform for seismic data interpolation[J]. Oil Geophysical Prospecting，2023，58(3)：617−625.

[3] SPITZ S. Seismic trace interpolation in the F–X domain[J]. Geophysics，1991，56(6)：785−794.

[4] CURRY W，SHAN Guojian. Interpolation of near offsets using multiples and prediction–error filters[J]. Geophysics，2010，75(6)：WB153−WB164.

[5] LIU Guochang，LI Chao，GUO Zhifeng，et al. Irregularly sampled seismic data reconstruction using multiscale multidirectional adaptive prediction–error filter[J]. IEEE Transactions on Geoscience and Remote Sensing，2019，57(5)：2909−2919.

[6] FOMEL S. Seismic reflection data interpolation with differential offset and shot continuation[J]. Geophysics，2003，68(2)：733−744.

[7] 辛可锋，王华忠，王成礼，等. 叠前地震数据的规则化[J]. 石油地球物理勘探，2002，37(4)：311−317.

XIN Kefeng，WANG Huazhong，WANG Chengli，et al. Regularization of pre–stack seismic data[J]. Oil Geophysical Prospecting，2002，37(4)：311−317.

[8] 周舟. 基于多道奇异谱分析方法的地震数据重建[D]. 北京：中国地质大学(北京)，2014.

ZHOU Zhou. Seismic data reconstruction based on multichannel singular spectrum analysis method[D]. Beijing：China University of Geosciences (Beijing)，2014.

[9] CADZOW J A. Signal enhancement：A composite property mapping algorithm[J]. IEEE Transactions on Acoustics，Speech，and Signal Processing，1988，36(1)：49−62.

[10] OROPEZA V，SACCHI M. Simultaneous seismic data denoising and reconstruction via multichannel singular spectrum analysis[J]. Geophysics，2011，76(3)：V25−V32.

[11] ABMA R，KABIR N. 3D interpolation of irregular data with a POCS algorithm[J]. Geophysics，2006，71(6)：E91−E97.

[12] 王敏玲，吴祺铭，王洪华，等. 基于区域阈值模型的地震信号凸集投影高效重建方法[J]. 石油地球物理勘探，2024，59(2)：195−205.

WANG Minling，WU Qiming，WANG Honghua，et al. POCS high–efficient reconstruction method of seismic signals based on regional threshold model[J]. Oil Geophysical Prospecting，2024，59(2)：195−205.

[13] 刘洋，SERGEY F，刘财，等. 高阶seislet变换及其在随机噪声消除中的应用[J]. 地球物理学报，2009，52(8)：2142−2151.

LIU Yang，SERGEY F，LIU Cai，et al. High–order seislet transform and its application of random noise attenuation[J]. Chinese Journal of Geophysics，2009，52(8)：2142−2151.

[14] CANDES E J，ROMBERG J K. Signal recovery from random projections[C]//Computational Imaging Ⅲ. San Jose：SPIE，2005：76.

[15] 刘国昌，陈小宏，郭志峰，等. 基于Curvelet变换的缺失地震数据插值方法[J]. 石油地球物理勘探，2011，46(2)：237−246.

LIU Guochang，CHEN Xiaohong，GUO Zhifeng，et al. Missing seismic data rebuilding by interpolation based on Curvelet transform[J]. Oil Geophysical Prospecting，2011，46(2)：237−246.

[16] 王汉闯，陶春辉，陈生昌，等. 基于稀疏约束的地震数据高效采集方法理论研究[J]. 地球物理学报，2016，59(11)：4246−4265.

WANG Hanchuang，TAO Chunhui，CHEN Shengchang，et al. Study on highly efficient seismic data acquisition method and theory based on sparsity constraint[J]. Chinese Journal of Geophysics，2016，59(11)：4246−4265.

[17] 温睿，刘国昌，冉扬. 压缩感知地震数据重建中的三个关键因素分析[J]. 石油地球物理勘探，2018，53(4)：682−693.

WEN Rui，LIU Guochang，RAN Yang. Three key factors in seismic data reconstruction based on compressive sensing[J]. Oil Geophysical Prospecting，2018，53(4)：682−693.

[18] WANG Yanfei，CAO Jingjie，YANG Changchun. Recovery of seismic wavefields based on compressive sensing by an l₁–norm constrained trust region method and the piecewise random subsampling[J]. Geophysical Journal International，2011，187(1)：199−213.

[19] 王本锋，陈小宏，李景叶，等. POCS联合改进的Jitter采样理论曲波域地震数据重建[J]. 石油地球物理勘探，2015，50(1)：20−28.

WANG Benfeng，CHEN Xiaohong，LI Jingye，et al. Seismic data reconstruction based on POCS and improved Jittered sampling in the curvelet domain[J]. Oil Geophysical Prospecting，2015，50(1)：20−28.

[20] 欧炳霖，曾同生，柳天成，等. 基于Huber–U–Net网络的地震数据重建与去噪[J]. 地球物理学进展，2023，38(6)：2540−2552.

OU Binglin，ZENG Tongsheng，LIU Tiancheng，et al. Seismic data reconstruction and de–noising based on Huber–U–Net network[J]. Progress in Geophysics，2023，38(6)：2540−2552.

[21] 徐凯军，卢炎，王大勇，等. 基于深度学习的大地电磁二维反演研究[J]. 石油地球物理勘探，2024，59(5)：1174−1183.

XU Kaijun，LU Yan，WANG Dayong，et al. Two–dimensional magnetotelluric inversion based on deep learning[J]. Oil Geophysical Prospecting，2024，59(5)：1174−1183.

[22] 马国庆，王泽坤，李丽丽. 基于自注意力机制深度学习的重磁数据网格化和滤波方法[J]. 石油地球物理勘探，2022，57(1)：34−42.

MA Guoqing，WANG Zekun，LI Lili. Gridding and filtering method of gravity and magnetic data based on self–attention deep learning[J]. Oil Geophysical Prospecting，2022，57(1)：34−42.

[23] 岳静楠，李志明. 基于卷积字典学习网络的地震数据重建[J]. 工程地球物理学报，2023，20(3)：383−392.

YUE Jingnan，LI Zhiming. Seismic data reconstruction based on convolutional dictionary learning network[J]. Chinese Journal of Engineering Geophysics，2023，20(3)：383−392.

[24] 曹静杰，高康富，许银坡，等. 基于一种注意力机制U–Net的地震数据去噪方法[J]. 石油地球物理勘探，2024，59(4)：724−735.

CAO Jingjie，GAO Kangfu，XU Yinpo，et al. Seismic data de–noising method based on an attention mechanism U–Net[J]. Oil Geophysical Prospecting，2024，59(4)：724−735.

[25] ZHU Huiyu，SUN Mengyao，FU Haohuan，et al. Training a seismogram discriminator based on ResNet[J]. IEEE Transactions on Geoscience and Remote Sensing，2021，59(8)：7076−7085.

[26] WANG Benfeng，ZHANG Ning，LU Wenkai，et al. Deep–learning–based seismic data interpolation：A preliminary result[J]. Geophysics，2019，84(1)：V11−V20.

[27] ZHOU Zongwei，SIDDIQUEE M M R，TAJBAKHSH N，et al. UNet++：A nested U–Net architecture for medical image segmentation[C]//Deep Learning in Medical Image Analysis and Multimodal Learning for Clinical Decision Support. Cham：Springer，2018：3–11.

[28] 李卿武，王兴建，张永恒，等. 基于3D U–Net++卷积神经网络的断层识别方法及应用[J]. 物探化探计算技术，2024，46(3)：284−291.

LI Qingwu，WANG Xingjian，ZHANG Yongheng，et al. Fault recognition method and application based on 3D U–Net++ convolution neural network[J]. Computing Techniques for Geophysical and Geochemical Exploration，2024，46(3)：284−291.

[29] 李振轩，黄敏儿，高飞，等. 基于U–Net、U–Net++和Attention–U–Net网络的遥感影像水体提取[J]. 测绘通报，2024(8)：26−30.

LI Zhenxuan，HUANG Miner，GAO Fei，et al. Remote sensing image water body extraction based on U–Net，U–Net++ and Attention–U–Net networks[J]. Bulletin of Surveying and Mapping，2024(8)：26−30.

[30] HUANG He，WANG Tengfei，CHENG Jiubing，et al. Self–supervised deep learning to reconstruct seismic data with consecutively missing traces[J]. IEEE Transactions on Geoscience and Remote Sensing，2022，60：5911514.

[31] JIN Xin，XIE Yanping，WEI Xiushen，et al. Delving deep into spatial pooling for squeeze–and–excitation networks[J]. Pattern Recognition，2022，121：108159.

[32] LI Haifeng，QIU Kaijian，CHEN Li，et al. SCAttNet：Semantic segmentation network with spatial and channel attention mechanism for high–resolution remote sensing images[J]. IEEE Geoscience and Remote Sensing Letters，2021，18(5)：905−909.

[33] HUANG Guoheng，ZHU Junwen，LI Jiajian，et al. Channel–attention U–Net：Channel attention mechanism for semantic segmentation of esophagus and esophageal cancer[J]. IEEE Access，2020，8：122798−122810.

[34] WOO S，PARK J，LEE J Y，et al. CBAM：Convolutional block attention module[C]//Computer Vision–ECCV 2018. Cham：Springer，2018：3–19.

[35] 杨润湉，马强，王志宝，等. 基于多尺度注意力UNet++的地震层位识别方法[J]. 石油物探，2025，64(2)：315−327.

YANG Runtian，MA Qiang，WANG Zhibao，et al. A seismic horizon identification method based on multi–scale attention UNet++[J]. Geophysical Prospecting for Petroleum，2025，64(2)：315−327.

[36] SEN Yang，WANG Zhenmin，SONG Wenlong，et al. The segmentation method of road surface covering objects based on CBAM UNet++[J]. IEEE Transactions on Emerging Topics in Computational Intelligence，2025，9(2)：1924−1933.

[37] WANG Junxin，ZHANG Qintong，XIE Hao，et al. Enhanced dual–channel model–based with improved UNet++ network for landslide monitoring and region extraction in remote sensing images[J]. Remote Sensing，2024，16(16)：2990.

[38] 张永洪，席梦丹. 带洞型U–Net++网络在遥感影像中建筑物的提取方法[J]. 测绘地理信息，2021，46(增刊1)：82−86.

ZHANG Yonghong，XI Mengdan. A method for extracting buildings from remote sensing images with hole U–Net++ network[J]. Journal of Geomatics，2021，46(Sup.1)：82−86.