A pre-trained dictionary learning framework combined with projections onto convex sets for seismic data reconstruction and denoising

Abstract

The dictionary learning approach has proven effective in seismic data denoising and interpolation. Its core advantage lies in the ability to continuously update the initial dictionary, thereby adapting to the complex structural characteristics of seismic data. However, many existing implementations rely on predefined transforms (e.g., discrete cosine transform) for dictionary initialization. These fixed, data-agnostic bases often fail to fully capture the unique features of seismic signals, which may compromise the sparsity and fidelity of signal representation. Such a limitation can significantly degrade the performance of tasks requiring high-precision reconstruction or noise attenuation. To address this issue, we propose an innovative dictionary learning framework based on a variational sparse representation model. Specifically, this framework first extracts small data patches from arbitrary locations in seismic data, and then constructs a pre-training dataset using a windowing algorithm to preserve fine-grained data features. This process yields an initial dictionary that inherently encodes the intrinsic characteristics of the input seismic data. Subsequently, the initial dictionary is separately refined and updated through the K-singular value decomposition (K-SVD) and sequential generalization of K-means (SGK) algorithms, resulting in an optimized dictionary with more accurate and data-adaptive features. In addition, we integrate a multi-iteration projections onto convex sets algorithm to compensate for missing data features, ultimately achieving high-precision seismic data interpolation and noise attenuation. Numerical experiments demonstrate that the proposed methods(variational SGK and variational K-SVD) outperform the conventional K-SVD and SGK algorithms in both interpolation accuracy and denoising performance.

Keywords

K-singular value decomposition / Sequential generalization of K-means / Variational sparse representation / Projections onto convex sets / Seismic data reconstruction and denoising

Acknowledgments

None.

Funding

This research was supported by the Open Fund of the National Engineering Research Center for Computer Software for Oil & Gas Exploration (grant no. DFWT-ZYRJ-2024-JS-83).

Conflict of interest

The authors declare no conflicts of interest.

Author contributions

Conceptualization: Jianlei Zhang, Min Bai, Xilin Qin, Baobin Wang, Zhen Zou

Formal analysis: Min Bai, Xilin Qin

Investigation: Jianlei Zhang, Bo Yang

Methodology: Jianlei Zhang, Bo Yang

Writing-original draft: Bo Yang

Writing-review & editing: Bo Yang, Boyuan Lv

Availability of data

The data generated or analyzed during this study are available from the corresponding author upon reasonable request.

References

[1] Canning A, Gardner GHF. Regularizing 3-D data sets with DMO. Geophysics. 1996; 61:1103-1114. doi: 10.1190/1.1444031

[2] Chemingui N, Biondi B. Handling the irregular geometry in wide-azimuth surveys. SEG Int Expo Annu Meet. 1996; 10: 32-35. doi: 10.1190/1.1826634

[3] Spitz S. Seismic trace interpolation in the F-X domain. Geophysics. 1991; 56:785-794. doi: 10.1190/1.1443096

[4] Li C, Liu G, Hao Z, Zu S, Mi F, Chen X. Multidimensional seismic data reconstruction using frequency-domain adaptive prediction-error filter. IEEE Trans Geosci Remote Sens. 2018; 56:2328-2336. doi: 10.1109/tgrs.2017.2778196

[5] Gao J, Sacchi MD, Chen X. A fast reduced-rank interpolation method for prestack seismic volumes that depend on four spatial dimensions. Geophysics. 2012;78:V21-V30. doi: 10.1190/geo2012-0038.1

[6] Zhang D, Zhou Y, Chen H, Chen W, Zu S, Chen Y. Hybrid rank-sparsity constraint model for simultaneous reconstruction and denoising of 3D seismic data. Geophysics. 2017;82:V351-V367. doi: 10.1190/geo2016-0557.1

[7] Gao J, Stanton A, Naghizadeh M, Sacchi MD, Chen X. Convergence improvement and noise attenuation considerations for beyond alias projection onto convex sets reconstruction. Geophys Prospect. 2013; 61:138-151. doi: 10.1111/j.1365-2478.2012.01103.x

[8] Trad DO, Ulrych TJ, Sacchi MD. Accurate interpolation with high-resolution time-variant radon transforms. Geophysics. 2002; 67:644-656. doi: 10.1190/1.1468626

[9] Hennenfent G, Fenelon L, Herrmann FJ. Nonequispaced curvelet transform for seismic data reconstruction: A sparsity-promoting approach. Geophysics. 2010;75:WB203-WB210. doi: 10.1190/1.3494032

[10] Fomel S, Liu Y. Seislet transform and seislet frame. Geophysics. 2010;75:V25-V38. doi: 10.1190/1.3380591

[11] Gan S, Wang S, Chen Y, Zhang Y, Jin Z. Dealiased seismic data interpolation using seislet transform with low-frequency constraint. IEEE Geosci Remote Sens Lett. 2015; 12: 2150-2154. doi: 10.1109/lgrs.2015.2453119

[12] Li H, Han L, Zhang L, Jia S. Seismic data reconstruction using dictionary learning and the alternating direction method of multipliers. Geophys Prospect Pet. 2019; 58(3):419-426. doi: 10.3969/j.issn.1000-1441.2019.03.011

[13] Wang B, Lu W, Chen X, Wang Z. Efficient seismic data interpolation using three-dimensional curvelet transform in the frequency domain. Geophys Prospect Pet. 2018; 57(1):65-71. doi: 10.3969/j.issn.1000-1441.2018.01.009

[14] Zhu L, Liu E, McClellan JH. Joint seismic data denoising and interpolation with double-sparsity dictionary learning. J Geophys Eng. 2017; 14(4):802-810. doi: 10.1088/1742-2140/aa6491

[15] Gelis LE, Gómez JL, Velis DR. Structured Noise Removal with Dictionary Learning in Seismic Data: The Effect of Atom Filtering. In: Proceedings of the IEEE International Workshop on Remote Sensing with Intelligent Processing (RPIC). Salvador, Brazil: IEEE; 2019. p. 204-208. doi: 10.1109/rpic.2019.8882163

[16] Wang W, Yang J, Huang J, Li Z, Sun M. Outlier denoising using a novel statistics-based mask strategy for compressive sensing. Remote Sens. 2023; 15(2):447. doi: 10.3390/rs15020447

[17] Cai J, Ji H, Shen Z, Ye G. Data-driven tight frame construction and image denoising. Appl Comput Harmon Anal. 2014; 37:89-105. doi: 10.1016/j.acha.2013.10.001

[18] Zhou Z, Bai M, Wu J, Cui Y. Coherent noise attenuation by kurtosis-guided adaptive dictionary learning based on variational sparse representation. IEEE Trans Geosci Remote Sens. 2023; 61:1-10. doi: 10.1109/tgrs.2023.3286791

[19] Wang W, Yang J, Li Z, Huang J, Zhao C. Seismic data denoising using a new framework of FABEMD-based dictionary learning. IEEE Trans Geosci Remote Sens. 2024; 62:1-9. doi: 10.1109/tgrs.2024.3396459

[20] Oropeza V, Sacchi M. Simultaneous seismic data denoising and reconstruction via multichannel singular spectrum analysis (MSSA). Geophysics. 2011;76:V25-V32. doi: 10.1190/1.3552706

[21] Naghizadeh M. Seismic data interpolation and denoising in the frequency-wavenumber domain. Geophysics. 2012; 77:71-80. doi: 10.1190/geo2011-0172.1

[22] Siahsar MA, Gholtashi S, Abolghasemi V, Chen Y. Simultaneous denoising and interpolation of 2D seismic data using data-driven non-negative dictionary learning. Signal Process. 2017; 141:309-321.doi: 10.1016/j.sigpro.2017.06.017

[23] Yu S, Ma J, Zhang X, Sacchi MD. Interpolation and denoising of high-dimensional seismic data by learning a tight frame. Geophysics. 2015;80:V119-V132. doi: 10.1190/geo2014-0396.1

[24] Turquais P, Asgedom EG, Söllner W, Gelius L. Parabolic dictionary learning for seismic wavefield reconstruction across the streamers. Geophysics. 2018;83:V263-V282. doi: 10.1190/geo2017-0694.1

[25] Beckouche S, Ma J. Simultaneous dictionary learning and denoising for seismic data. Geophysics. 2014; 79(3):A27-A31. doi: 10.1190/geo2013-0382.1

[26] Turquais P, Asgedom EG, Söllner W. A method of combining coherence-constrained sparse coding and dictionary learning for denoising. Geophysics. 2017; 82(3):V137-V148. doi: 10.1190/geo2016-0164.1

[27] Siahsar MA, Gholtashi S, Torshizi EO, Chen W, Chen Y. Simultaneous denoising and interpolation of 3-D seismic data via damped data-driven optimal singular value shrinkage. IEEE Geosci Remote Sens Lett. 2017; 14: 1086-1090. doi: 10.1109/lgrs.2017.2697942

[28] Chen W, Saad OM, Oboué YASI, Yang L, Chen Y. Retrieving the leaked signals from noise using a fast dictionary-learning method. Geophysics. 2021; 87(1):V39-V49. doi: 10.1190/geo2021-0243.1

[29] Aharon M, Elad M, Bruckstein A. K-SVD: An algorithm for designing overcomplete dictionaries for sparse representation. IEEE Trans Signal Process. 2006; 54(11):4311-4322. doi: 10.1109/tsp.2006.881199

[30] Rubinstein R, Zibulevsky M, Elad M. Double sparsity: Learning sparse dictionaries for sparse signal approximation. IEEE Trans Signal Process. 2009; 58(3):1553-1564.doi: 10.1109/tsp.2009.2036477

[31] Wang B, Wu R, Chen X, Li J. Simultaneous seismic data interpolation and denoising with a new adaptive method based on dreamlet transform. Geophys J Int. 2015; 201(2):1182-1194.doi: 10.1093/gji/ggv072

[32] Almadani M, Waheed UB, Masood M, Chen Y. Dictionary learning with convolutional structure for seismic data denoising and interpolation. Geophysics. 2021; 86(5):V361-V374.doi: 10.1190/geo2019-0689.1

[33] Li C, Wen X, Liu X, Zu S. Simultaneous seismic data interpolation and denoising based on nonsubsampled contourlet transform integrating with two-step iterative log thresholding algorithm. IEEE Trans Geosci Remote Sens. 2022; 60:1-10.doi: 10.1109/tgrs.2022.3192531

[34] Kaplan ST, Sacchi MD, Ulrych TJ. Sparse coding for data-driven coherent and incoherent noise attenuation. In: SEG Technical Program Expanded Abstracts; 2009. p. 3327-3331. doi: 10.1190/1.3255551

[35] Wang H, Chen W, Zhang Q, Liu X, Zu S, Chen Y. Fast dictionary learning for high-dimensional seismic reconstruction. IEEE Trans Geosci Remote Sens. 2021; 59(8):7098-7108.doi: 10.1109/tgrs.2020.3030740