Adaptive and High-Precision Isosurface Meshes from CT Data
Lin Xue,Jialong Xu,Kai Ma,Zhaoxiang Li,Jingtao Wang
DOI: https://doi.org/10.1007/s10921-024-01102-8
2024-07-18
Journal of Nondestructive Evaluation
Abstract:This paper proposes a method for obtaining adaptive and high-precision surface meshes directly from Industrial computed tomography (ICT) projection data. Firstly, an adaptive volume octree is recursively constructed from top to bottom using a two-stage geometric error metric function. The CT values and gradient values at the nodes are computed using the Feldkamp–Davis–Kress (FDK) reconstruction algorithm and its derivatives, achieving sub-voxel precision. Next, feature vertices are calculated based on Quadratic error functions (QEFs), and a dual mesh is constructed. Finally, Hermite interpolation is used to determine the iso-surface vertices, and the Convex Contouring lookup table is employed to accurately extract the iso-surface contours, resulting in high-precision and crack-free surface meshes. Experimental results show that the surface meshes generated by the proposed method exhibit superior dimensional accuracy, form and position accuracy, and surface model accuracy compared to traditional methods, and the dimensional accuracy has been enhanced by approximately 10–30%.
materials science, characterization & testing