Stochastic Reconstruction of an Oolitic Limestone by Generative Adversarial Networks

Abstract

Stochastic image reconstruction is a key part of modern digital rock physics and material analysis that aims to create representative samples of microstructures for upsampling, upscaling and uncertainty quantification. We present new results of a method of three-dimensional stochastic image reconstruction based on generative adversarial neural networks (GANs). GANs are a family of unsupervised learning methods that require no a priori inference of the probability distribution associated with the training data. Thanks to the use of two convolutional neural networks, the discriminator and the generator, in the training phase, and only the generator in the simulation phase, GANs allow the sampling of large and realistic volumetric images. We apply a GAN-based workflow of training and image generation to an oolitic Ketton limestone micro-CT unsegmented gray-level dataset. Minkowski functionals calculated as a function of the segmentation threshold are compared between simulated and acquired images. Flow simulations are run on the segmented images, and effective permeability and velocity distributions of simulated flow are also compared. Results show that GANs allow a fast and accurate reconstruction of the evaluated image dataset. We discuss the performance of GANs in relation to other simulation techniques and stress the benefits resulting from the use of convolutional neural networks . We address a number of challenges involved in GANs, in particular the representation of the probability distribution associated with the training data.

Computed Tomography 3D Super-Resolution with Generative Adversarial Neural Networks: Implications on Unsaturated and Two-Phase Fluid Flow

“…The success of a lot of these SR applications is explained by the emergence of generative adversarial networks (GANs) [108] which are commonly known for their strength at generating realistic "fake" images [109]. These GANs have been used for geomaterial porosity simulation [110][111][112], but even though they have proven their capacity as super-resolution methods, they have not yet been applied to super-resolution of geological materials. In fact, neural networks have barely been used to solve this problem, except by Wang et al [84,85] and Shams et al [88].…”

Parametrization of Stochastic Inputs Using Generative Adversarial Networks With Application in Geology

“…In the domain of geology, a natural question is whether GANs can be directly applied as multipoint geostatistical simulators. This has been studied in a number of recent works [36,37,30]. The idea here is to use a single large training image and simply train a GAN model on patches of this image, instead of generating a dataset of realizations using an external multipoint geostatistical simulator.…”

“…There is a growing number of recent works in geology-related fields where GAN methods have been studied. In [36,37], GAN is used to generate images of porous media for image reconstruction. In [38], GAN is used in seismic inversion.…”

Microstructural Materials Design Via Deep Adversarial Learning Methodology

“…Different from VAE, GAN identifies the latent variables of data by training a generator-discriminator model pair in adversarial manner. In [33,34], GAN is used for reconstructing different types of microstructures, but their applications in computational materials design are unexplored. In this work, as illustrated in Figure 1, we apply a fully scalable GAN-based approach to determine the latent variables of a set of microstructures once its dimensionality is pre-specified.…”