Generation of ground truth images to validate micro-CT image-processing pipelines

Abstract

Digital rock technology and pore-scale physics have become increasingly relevant topics in a wide range of porous media with important applications in subsurface engineering. This technology relies heavily on images of pore space and pore-level fluid distribution determined by X-ray microcomputed tomography (micro-CT). Digital images of pore space (or pore-scale fluid distribution) are typically obtained as gray-level images that first need to be processed and segmented to obtain the binary images that uniquely represent rock and pore (including fluid phases). This processing step is not trivial. Rock complexity, image quality, noise, and other artifacts prohibit the use of a standard processing workflow. Instead, an array of strategies of increasing sophistication has been developed. Typical processing pipelines consist of filtering, segmentation, and postprocessing steps. For each step, various choices and different options exist. This makes selection and validation of an optimum processing pipeline difficult. Using Darcy-scale quantities as a benchmark is not a good option because of rock heterogeneity and different scales of observation. Here, we present a conceptual workflow where noisy images are derived from a ground truth by systematically including typical image artifacts and noise. Artifacts and noise are not simply added to the images. Instead, tomographic forward projection and reconstruction steps are used to incorporate the artifacts in a physically correct way. A proof of concept of this workflow is demonstrated by comparing seven different image-segmentation pipelines ranging from absolute thresholding to a machine-learning approach (Trainable Weka Segmentation). The Trainable Weka Segmentation showed the best performance of the tested methods.

Using nano-XRM and high-contrast imaging to inform micro-porosity permeability during Stokes-Brinkman single and two-phase flow simulations on micro-CT images

“…The resulting reconstructed image was first denoised using an edge preserving non-local means filter, then segmented using ZEISS Zen Intellesis machine learning based segmentation. Such a segmentation technique has been showed in quantitative benchmarks to be significantly more robust when dealing with such noisy and challenging images [34,35]. The resulting porosity observed within the image (41%) matched well with the inferred porosity of the 30µm x 30µm x 30µm region initially identified from the macroscopic 10µm voxel size image of the 10mm diameter core.…”

Digital core repository coupled with machine learning as a tool to classify and assess petrophysical rock properties

“…Chen and Zeng [10] demonstrated the capability of machine learning in rock facies classification from wireline log scalar attributes improving by feature augmentation. Compared to conventional image segmentation methods, machinelearning segmentation could come closer to the ground truth for determining the porosity from noisy MCT images (Berg et al [11]). Karimpouli and Tahmasebi [12] revealed that a CNN algorithm improves the accuracy of a segmentation comparing with a multiphase thresholding segmentation.…”

An Application of Deep Neural Networks for Segmentation of Microtomographic Images of Rock Samples

“…There are many industrial applications of DR [3]: estimation of fluid transport properties such as absolute and relative permeabilities [4][5][6]; assessment of enhanced oil recovery (EOR) methods [7,8]; calculation of elastic moduli and electrical conductivity [9]; and modelling of nuclear magnetic resonance (NMR) response [10]. Figure 1 shows a typical DR workflow [11]. The first step is image acquisition by X-ray microcomputed tomography (microCT) [12].…”