Intersection-free tetrahedral meshing from volumetric images

High-fidelity meshes are essential in many visualisation and simulation applications. In this study, we develop an automatic approach to generate intersection-free meshes from volumetric images. Due to the properties of octree-based dual contouring, the generated elements may intersect with each other. We need a robust method to generate quality intersection-free tetrahedral meshes with no topology ambiguity. In our algorithm, we first analyse each sign-change edge or interior edge to define an envelope, and then check whether self-intersections will appear. In the end, we choose appropriate meshing technique to create intersection-free meshes. Our main contribution lies in intersection-free tetrahedral meshing for single-material and multiple-material domains, with topology ambiguities resolved by handling hybrid grids. We have applied our method to two biomedical and one microstructure data-sets and obtained good results based on our quality metrics.