Topological characterization of microstructures from 3D digitized voxel data

Topological characterization of 3D voxel-based images is non-trivial. Since there are many scientific simulations that rely on digitized images consisting of voxel data, and usually need to be run in parallel, an accurate and efficient algorithm to characterize 3D microstructures in cubic lattice is crucial for research on topology. In this work, we present an algorithm that enables detailed topological characterization that is extendible to parallel computing while retaining its efficiency and robustness. With this algorithm, a 3D microstructure made up of voxels can be accurately described in terms of faces, edges and vertices and the network formed by them. This algorithm is applied to various data sets consisting of idealized, regular structures and to microstructures generated by grain growth simulations. Highly accurate characterization of topology has been achieved for all data sets.