«Automation and Informatization of the fuel and energy complex»

Multi-class segmentation of heterogeneous porous media using 3D U-Net and Minkowski functional fields

UDC: 004.8
DOI: -

Authors:

¹ National University of Oil and Gas "Gubkin University", Moscow, Russia

Keywords: 3D U-Net, multi-class segmentation, computational morphology, Minkowski functionals

Annotation:

The authors of the article propose a method for training a 3D U-Net neural network for multi-class segmentation of 3D digital core samples. The training is performed on the data segmented by calculating the fields of morphological descriptor – Minkowski functionals. Digital samples were generated using the Gaussian Random Field method: some volume regions were filled with grains of different radii. Then, Minkowski functionals fields were calculated for each sample. The segmentation problem was solved by applying the Gaussian Mixture Model to the Minkowski functionals fields, resulting in a large enough dataset to train a convolutional neural network with 3D U-Net architecture. The trained neural network demonstrated stable results on both synthetic and real data sets.

Bibliography:

1. Arsenyev-Obraztsov S. Generation of Petrophysical Parameters for Forecasting of Oil and Gas Deposits Development, Digital Core and Multi-scale // 2021 14th Int. Conf. Management of Large-scale System Development (MLSD), Moscow, Russian Federation, Sept. 27–29, 2021. – Institute of Electrical and Electronics Engineers Inc., 2021. – P. 9600135. – DOI: 10.1109/MLSD52249.2021.9600135
2. Arsen'ev-Obraztsov S.S., Volkov E.A. Klasterizatsiya tsifrovogo 3D-obraztsa kerna s ispol'zovaniem funktsionalov Minkovskogo // Avtomatizatsiya i informatizatsiya TEK. – 2023. – № 8(601). – S. 13–22. – DOI: 10.33285/2782-604X-2023-8(601)-13-22
3. Arsenyev-Obraztsov S.S., Volkov E.A., Plusch G.O. Proposals on 3D parallel edge-preserving filtration for x-ray tomographic digital images of porous medium core plugs // IOP Conf. Series: Materials Science and Engineering. – IOP Publishing, 2019. – Vol. 700, Isuue 1. 2nd Conf. of Computational Methods in Offshore Technology and First Conf. of Oil and Gas Technology (COTech & OGTech 2019), Nov. 27–29, 2019, Stavanger, Norway. – P. 012053. – DOI: 10.1088/1757-899X/700/1/012053
4. Arsen'ev-Obraztsov S.S., Volkov E.A. Realizatsiya diskretnogo preobrazovaniya Khartli na mnogoyadernykh uskoritelyakh // Avtomatizatsiya i informatizatsiya TEK. – 2023. – № 7(600). – S. 35–42. – DOI: 10.33285/2782-604X-2023-7(600)-35-42
5. Beifang Chen. A Simplified Elementary Proof of Hadwiger's Volume Theorem // Geometriae Dedicata. – 2004. – Vol. 105. – P. 107–120. – DOI: 10.1023/B:GEOM.0000024665.02286.46
6. Mecke K.R. Additivity, convexity, and beyond: applications of Minkowski functionals in statistical physics // Statistical Physics and Spatial Statistics: The art of analyzing and modeling spatial structures and pattern formation. – Berlin, Heidelberg: Springer Berlin Heidelberg, 2000. – P. 111–184.
7. Han Jiang, Arns C.H. Fast Fourier transform and support-shift techniques for pore-scale microstructure classification using additive morphological measures // Physical Review E. – 2020. – Vol. 101, Issue. 3. – P. 033302. – DOI: 10.1103/PhysRevE.101.033302
8. Downscaling-Based Segmentation for Unresolved Images of Highly Heterogeneous Granular Porous Samples / S. Korneev, Xiaofan Yang, J. Zachara [et al.] // Water Resources Research. – 2018. – Vol. 54, Issue 4. – P. 2871–2890. – DOI: 10.1002/2018WR022886
9. 3D U-Net: Learning Dense Volumetric Segmentation from Sparse Annotation / Ö. Çiçek, A. Abdulkadir, S.S. Lienkamp [et al.] // Medical Image Computing and Computer-Assisted Intervention – MICCAI 2016: 19th Int. Conf., Athens, Greece, Oct. 17–21, 2016. – Springer Int. Publishing, 2016. – Part II. – P. 424–432. – DOI: 10.48550/arXiv.1606.06650
10. Zomorodian A., Carlsson G. Computing Persistent Homology // Proceedings of the twentieth annual symposium on Computational geometry. – 2004. – P. 347–356. – DOI: 10.1145/997817.997870
11. Edelsbrunner H., Harer J. Persistent Homology – a Survey // Contemporary mathematics. – 2008. – Vol. 453, Issue 26. – P. 257–282. – DOI: 10.1090/conm/453/08802
12. Mecke K.R. Integral Geometry in Statistical Physics // Int. J. of Modern Physics B. – 1998. – Vol. 12, No. 09. – P. 861–899. – DOI: 10.1142/S0217979298000491
13. Klain D.A. A short proof of Hadwiger's characterization theorem // Mathematika. – 1995. – Vol. 42, Issue 2. – P. 329–339. – DOI: 10.1112/S0025579300014625