نشریه ژئومکانیک و ژئوانرژی

نشریه ژئومکانیک و ژئوانرژی

Three-Dimensional Reconstruction of Porous Media Images Using the Vox2Vox Model in Presence of Multimineral Segmentation Information

نوع مقاله : مقاله پژوهشی

نویسندگان
بهاره کشاورز 1
محسن مسیحی 2
مستانه حاجی پور شیرازی فرد 3
ابراهیم بی نیاز دلیجانی 3
1 دانشگاه آزاد اسلامی واحد علوم تحقیقات، دانشکده نفت و مهندسی شیمی، تهران
2 استاد مهندسی نفت دانشکده مهندسی شیمی نفت دانشگاه صنعتی شریف
3 دانشکده نفت و مهندسی شیمی، دانشگاه آزاد اسلامی واحد علوم و تحقیقات، تهران، ایران
10.22107/jpg.2025.436814.1226
چکیده
The evaluation of morphological, topological, statistical, and flow properties of porous media needs high-resolution images of porous media at the pore scale. However, direct access to high-quality tomographic images can be costly and impractical. Image reconstruction techniques offer a viable solution for obtaining visually realistic image data. Among these methods, deep learning-based approaches have gained significant attention from researchers. However, the utilization of petrographic information from porous media in training such models is rare.
In this study, we investigate the impact of incorporating multimineral segmentation information in the training of the VOX2VOX model. The data used in this model consists of 3D images of a sandstone reservoir along with segmented images into 5 classes, including macro-pores, clay, quartz, feldspar, and high-density minerals. Additionally, the Vox2Vox model has been trained with binary segmented images containing pores and solid phases to compare the effect of multimineral information on model and reconstructed images. Incorporating multi-mineral segmentation from the porous media significantly enhances the model's image reconstruction capabilities, as observed through comparisons of various dynamic and static features. Incorporating a five-class dataset has led the model to exhibit lower error at the outset of training, stabilizing after approximately 30 epochs, whereas this point for the model based on the two-class dataset is around 60 epochs. Furthermore, the comparison parameters for image quality, specifically SSIM Score: 0.95, MSE Score: 0.00013, and PSNR Score: 38.93, are observed for the first model, while for the second model, they are SSIM Score: 0.89, MSE Score: 0.00112, and PSNR Score: 29.49. The original image has a porosity of 0.229, which increases to 0.233 and 0.244 in the reconstructed images from the models based on five-class and two-class datasets, respectively. Additionally, the graphs of other parameters also demonstrate the superiority of the model based on the five-class dataset.
کلیدواژه‌ها
Deep Learning
Image Reconstruction
Multimineral Segmentation
Vox2Vox model

عنوان مقاله English

Three-Dimensional Reconstruction of Porous Media Images Using the Vox2Vox Model in Presence of Multi mineral Segmentation Information

نویسندگان English

Bahareh Keshavarz 1
Mohsen Masihi 2
Mastaneh Hajipour Shirazi Fard 3
Ebrahim Biniaz Delijani 3
1 Faculty of Petroleum and Chemical Engineering, Islamic Azad University, Science and Research Branch, Tehran
2 Professor of Petroleum Engineering, Department of Chemical and Petroleum Engineering, Sharif University of Technology
3 Faculty of Petroleum and Chemical Engineering, Islamic Azad University Science and Research Branch, , Tehran, Iran
چکیده English

The evaluation of morphological, topological, statistical, and flow properties of porous media needs high-resolution images of porous media at the pore scale. However, direct access to high-quality tomographic images can be costly and impractical. Image reconstruction techniques offer a viable solution for obtaining visually realistic image data. Among these methods, deep learning-based approaches have gained significant attention from researchers. However, the utilization of petrographic information from porous media in training such models is rare.
In this study, we investigate the impact of incorporating multimineral segmentation information in the training of the VOX2VOX model. The data used in this model consists of 3D images of a sandstone reservoir along with segmented images into 5 classes, including macro-pores, clay, quartz, feldspar, and high-density minerals. Additionally, the Vox2Vox model has been trained with binary segmented images containing pores and solid phases to compare the effect of multimineral information on model and reconstructed images. Incorporating multi-mineral segmentation from the porous media significantly enhances the model's image reconstruction capabilities, as observed through comparisons of various dynamic and static features. Incorporating a five-class dataset has led the model to exhibit lower error at the outset of training, stabilizing after approximately 30 epochs, whereas this point for the model based on the two-class dataset is around 60 epochs. Furthermore, the comparison parameters for image quality, specifically SSIM Score: 0.95, MSE Score: 0.00013, and PSNR Score: 38.93, are observed for the first model, while for the second model, they are SSIM Score: 0.89, MSE Score: 0.00112, and PSNR Score: 29.49. The original image has a porosity of 0.229, which increases to 0.233 and 0.244 in the reconstructed images from the models based on five-class and two-class datasets, respectively. Additionally, the graphs of other parameters also demonstrate the superiority of the model based on the five-class dataset.

کلیدواژه‌ها English

Deep Learning
Image Reconstruction
porous media
Multi mineral Segmentation
Vox2Vox model
[1] S. Torquato, “Random heterogeneous materials: microstructure and macroscopic properties,” p. 701, 2002.
[2] P. Tahmasebi, “Accurate modeling and evaluation of microstructures in complex materials,” Phys Rev E, vol. 97, no. 2–1, Feb. 2018, doi: 10.1103/PHYSREVE.97.023307.
[3] G. T. Herman, “Fundamentals of Computerized Tomography,” 2009, doi: 10.1007/978-1-84628-723-7.
[4] K. J. Lange, H. Carlsson, I. Stewart, P. C. Sui, R. Herring, and N. Djilali, “PEM fuel cell CL characterization using a standalone FIB and SEM: Experiments and simulation,” Electrochim Acta, vol. 85, pp. 322–331, Dec. 2012, doi: 10.1016/J.ELECTACTA.2012.08.082.
[5] J. Baruchel et al., “Advances in synchrotron hard X-ray based imaging,” C R Phys, vol. 9, no. 5–6, pp. 624–641, Jun. 2008, doi: 10.1016/J.CRHY.2007.08.003.
[6] P. Mostaghimi, M. J. Blunt, and B. Bijeljic, “Computations of Absolute Permeability on Micro-CT Images,” Math Geosci, vol. 45, no. 1, pp. 103–125, Jan. 2013, doi: 10.1007/S11004-012-9431-4/METRICS.
[7] S. Kamrava, P. Tahmasebi, and M. Sahimi, “Enhancing images of shale formations by a hybrid stochastic and deep learning algorithm,” Neural Networks, vol. 118, pp. 310–320, Oct. 2019, doi: 10.1016/J.NEUNET.2019.07.009.
[8] P. Tahmasebi, “Nanoscale and multiresolution models for shale samples,” Fuel, vol. 217, pp. 218–225, Apr. 2018, doi: 10.1016/J.FUEL.2017.12.107.
[9] F. Javadpour, D. Fisher, and M. Unsworth, “Nanoscale gas flow in shale gas sediments,” Journal of Canadian Petroleum Technology, vol. 46, no. 10, pp. 55–61, Oct. 2007, doi: 10.2118/07-10-06/180817/NANOSCALE-GAS-FLOW-IN-SHALE-GAS-SEDIMENTS.
[10] J. L. Kulikowski, “Computer-Aided Porous Materials Description Based on Morphological Spectra,” Research & Development in Material Science, vol. 11, no. 4, Sep. 2019, doi: 10.31031/RDMS.2019.11.000769.
[11] Q. Yang, J. Yao, Z. Huang, G. Zhu, L. Liu, and W. Song, “Pore-scale investigation of petro-physical fluid behaviors based on multiphase SPH method,” J Pet Sci Eng, vol. 192, p. 107238, Sep. 2020, doi: 10.1016/J.PETROL.2020.107238.
[12] A. Rabbani, S. Jamshidi, and S. Salehi, “An automated simple algorithm for realistic pore network extraction from micro-tomography images,” J Pet Sci Eng, vol. 123, pp. 164–171, Nov. 2014, doi: 10.1016/J.PETROL.2014.08.020.
[13] J. Feng, X. He, Q. Teng, C. Ren, H. Chen, and Y. Li, “Accurate and Fast reconstruction of Porous Media from Extremely Limited Information Using Conditional Generative Adversarial Network,” Phys Rev E, vol. 100, no. 3, Apr. 2019, doi: 10.1103/PhysRevE.100.033308.
[14] S. Kamrava, M. Sahimi, and P. Tahmasebi, “Quantifying accuracy of stochastic methods of reconstructing complex materials by deep learning,” Phys Rev E, vol. 101, no. 4, p. 043301, Apr. 2020, doi: 10.1103/PHYSREVE.101.043301/FIGURES/9/MEDIUM.
[15] S. Kamrava, P. Tahmasebi, and M. Sahimi, “Linking Morphology of Porous Media to Their Macroscopic Permeability by Deep Learning,” Transp Porous Media, vol. 131, no. 2, pp. 427–448, Jan. 2020, doi: 10.1007/S11242-019-01352-5/METRICS.
[16] I. J. Goodfellow et al., “Generative Adversarial Networks,” Jun. 2014, [Online]. Available: http://arxiv.org/abs/1406.2661.
[17] P. Isola, J. Y. Zhu, T. Zhou, and A. A. Efros, “Image-to-Image Translation with Conditional Adversarial Networks,” Proceedings - 30th IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017, vol. 2017-January, pp. 5967–5976, Nov. 2016, doi: 10.1109/CVPR.2017.632.
[18] W. Pan, C. Torres-Verdin, and M. J. Pyrcz, “Stochastic Pix2pix: A New Machine Learning Method for Geophysical and Well Conditioning of Rule-Based Channel Reservoir Models,” Nat Resour Res 30, 1319–1345 (2021). https://doi.org/10.1007/s11053-020-09778-1.
[19] G. Buono, S. Caliro, G. Macedonio et al. Exploring microstructure and petrophysical properties of microporous volcanic rocks through 3D multiscale and super-resolution imaging. Sci Rep 13, 6651 (2023). https://doi.org/10.1038/s41598-023-33687-x.
[20] T.I. Anderson, B.Vega, J. McKinzie et al. 2D-to-3D image translation of complex nanoporous volumes using generative networks. Sci Rep 11, 20768 (2021). https://doi.org/10.1038/s41598-021-00080-5
[21] R. Shams, M. Masihi, R. B. Boozarjomehry, and M. J. Blunt, “Coupled generative adversarial and auto-encoder neural networks to reconstruct three-dimensional multi-scale porous media,” J Pet Sci Eng, vol. 186, p. 106794, Mar. 2020, doi: 10.1016/J.PETROL.2019.106794.
[22] Y. Da Wang, M. Shabaninejad, R. T. Armstrong, and P. Mostaghimi, “Deep neural networks for improving physical accuracy of 2D and 3D multi-mineral segmentation of rock micro-CT images,” Appl Soft Comput, vol. 104, Jun. 2021, doi: 10.1016/j.asoc.2021.107185.
[23] F. Zhang, Q. Teng, H. Chen, X. He, and X. Dong, “Slice-to-voxel stochastic reconstructions on porous media with hybrid deep generative model,” Comput Mater Sci, vol. 186, p. 110018, Jan. 2021, doi: 10.1016/J.COMMATSCI.2020.110018.
[25] J. Feng, Q. Teng, X. He, and X. Wu, “Accelerating multi-point statistics reconstruction method for porous media via deep learning,” Acta Mater, vol. 159, pp. 296–308, Oct. 2018, doi: 10.1016/J.ACTAMAT.2018.08.026.
[26] J. Feng, X. He, Q. Teng, C. Ren, H. Chen, and Y. Li, “Reconstruction of porous media from extremely limited information using conditional generative adversarial networks,” Phys Rev E, vol. 100, no. 3, p. 033308, Sep. 2019, doi: 10.1103/PHYSREVE.100.033308/FIGURES/22/MEDIUM.
[27] R. Shams, M. Masihi, R. B. Boozarjomehry, and M. J. Blunt, “A hybrid of statistical and conditional generative adversarial neural network approaches for reconstruction of 3D porous media (ST-CGAN),” Adv Water Resour, vol. 158, p. 104064, Dec. 2021, doi: 10.1016/J.ADVWATRES.2021.104064.
[28] K. He, X. Zhang, S. Ren, and J. Sun, “Deep Residual Learning for Image Recognition,” Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, vol. 2016-December, pp. 770–778, Dec. 2015, doi: 10.1109/CVPR.2016.90.
[29] M. D. Cirillo, D. Abramian, and A. Eklund, “Vox2Vox: 3D-GAN for Brain Tumour Segmentation,” Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 12658 LNCS, pp. 274–284, Mar. 2020, doi: 10.1007/978-3-030-72084-1_25. 
[30] D. Ulyanov, A. Vedaldi, and V. Lempitsky, “Instance Normalization: The Missing Ingredient for Fast Stylization,” Jul. 2016, Accessed: Aug. 06, 2023. [Online]. Available: https://arxiv.org/abs/1607.08022v3
[31] K. He, X. Zhang, S. Ren, and J. Sun, “Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification,” Feb. 2015, [Online]. Available: http://arxiv.org/a
نشریه ژئومکانیک و ژئوانرژی
دوره 7، شماره 4
زمستان 1403
صفحه 1-13