Efficient Boundary Surface Reconstruction from Heterogeneous Volumetric Data Via Tri-Prism Decomposition

Ming Wang,Jie-Qing Feng,Wei Chen

DOI: https://doi.org/10.1016/j.cag.2013.10.034

IF: 1.821

2014-01-01

Computers & Graphics

Abstract:We propose a novel and efficient approach for extracting the boundary surfaces from heterogeneous volumetric data in one pass. Each homogeneous material component is surrounded by a boundary surface, which is composed of piecewise 2-manifold meshes. The key idea is to subdivide each cubical voxel into two tri-prism voxels and to construct the boundary surfaces in a dimension-ascending (DA) way, i.e., from points to lines and then to faces. The extracted boundary surfaces can fully isolate the homogeneous material components, and the information on intersections between boundary surfaces can be explicitly retrieved. The surface reconstruction process can be accomplished efficiently by adopting a case table. The proposed approach is independent of the number of material types employed. Additionally, a new case index encoding approach is proposed to encode all possible cases in a heterogeneous tri-prism voxel that can verify the proposed DA approach in an exhaustive enumeration manner. The experimental results demonstrate that our approach can accurately and efficiently generate a boundary representation of heterogeneous volumetric data.