Three-Dimensional Structural Imaging of Rock Components and Methods for Component Segmentation and Extraction

Abstract

The variations in the internal composition and microstructure of rocks result in uncertainty in their macroscopic mechanical properties. To determine the internal composition and microscopic three-dimensional (3D) structure of a rock specimen accurately, the analog computed tomography (CT) signal from a sample was digitized and the x-ray absorption values (CT values) were converted into a two-dimensional digital matrix. The filtered backprojection algorithm was then used to reconstruct the 3D projection of the microscopic components of the rock. The nonlocal means algorithm was then used to correct for beam hardening, thereby improving the resolution of the boundaries of the microstructures. Based on the improved Otsu algorithm, the reconstructed 3D image of the rock was then segmented to extract the pores, cement, and mineral particles in the sandstone. In this way, the geometrical shape and spatial distribution of these components inside the rock could be accurately obtained. This approach is important because it can be used to realize 3D microscopic geometric models of rock samples, characterize rock geometry models quantitatively, and support accurate numerical simulations of physical models.