Image registration of 2D optical thin sections in a 3D porous medium: Application to a Berea sandstone digital rock image

TL;DR

The paper addresses the challenge of aligning a 2D thin-section image within a 3D CT volume to enable accurate multimodal analyses in digital rock physics. It introduces a four-step registration workflow that rotates the 3D volume and uses template matching with normalized cross-correlation, optimized by differential evolution to minimize $f(\alpha,\beta,\gamma,s) = - \max_{u,v} \text{NCC}(u,v \mid I_s, T)$. Validation on synthetic data shows angle errors below $0.001^\circ$ and a near-unique alignment with a peak correlation near 1, and application to Berea sandstone yields a structural similarity index (SSIM) of $0.990$. The paired analyses reveal porosity increases by about 50% in the thin section (0.232 vs 0.154) and mean pore size decreases (7.52 vs 18.80 $\mu$m), while bulk and shear moduli are reduced by $25\%$ and $30\%$, illustrating how multimodal registration can correct CT biases and support data-driven upscaling and super-resolution in digital rock physics.

Abstract

This study proposes a systematic image registration approach to align 2D optical thin-section images within a 3D digital rock volume. Using template image matching with differential evolution optimization, we identify the most similar 2D plane in 3D. The method is validated on a synthetic porous medium, achieving exact registration, and applied to Berea sandstone, where it achieves a structural similarity index (SSIM) of 0.990. With the registered images, we explore upscaling properties based on paired multimodal images, focusing on pore characteristics and effective elastic moduli. The thin-section image reveals 50 % more porosity and submicron pores than the registered CT plane. In addition, bulk and shear moduli from thin sections are 25 % and 30 % lower, respectively, than those derived from CT images. Beyond numerical comparisons, thin sections provide additional geological insights, including cementation, mineral phases, and weathering effects, which are not clear in CT images. This study demonstrates the potential of multimodal image registration to improve computed rock properties in digital rock physics by integrating complementary imaging modalities.

Figures (11)

• Figure 1: Thin section preparation process
• Figure 2: Workflow for 2D image registration within a 3D CT volume via optimization. The upper section represents the process of applying rotations to the 3D CT scan, extracting a 2D slice, and computing similarity with the template image. The lower section shows the differential evolution optimization process, which iteratively updates the transformation parameters to minimize the objective function $f$, derived from the similarity evaluation in the upper section.
• Figure 3: Validation of the 2D image registration approach in a 3D synthetic porous medium. (left) The initial 3D sphere pack porous medium. (middle) A known 2D plane selected as the ground truth (template) for the image registration (green). (right) The registered 2D image after image registration (blue), overlaid with the ground truth for validation.
• Figure 4: Validation case: comparison between the registered image extracted from the 3D volume ($I_s$) and the ground truth template ($T$).
• Figure 5: Validation case: normalized correlation coefficient maps between the 2D plane from 3D image ($I_s$) and ground truth template ($T$) in the Figure \ref{['fig:validation_superimposed']}