🤖 AI Summary
Reconstructing high-fidelity, topologically closed meshes from sparse or constrained 3D point clouds—especially those acquired from narrow cavities—remains challenging. This paper proposes a novel meshing method based on physically simulated flexible foil deformation. It is the first to integrate dynamic elasticity modeling into point cloud meshing, synergistically combining pressure-driven surface deformation, implicit surface initialization, spatially constrained optimization, and adaptive vertex snapping. This framework unifies geometric fidelity with physical plausibility while automatically generating watertight, topologically consistent surfaces. Extensive evaluation on diverse complex cavity point clouds demonstrates that our method reduces boundary fitting error by 37% compared to Poisson surface reconstruction and ball-pivoting, achieves 100% topological completeness, and significantly improves both accuracy and robustness of closed mesh generation.
📝 Abstract
We propose a method for constructing high-quality, closed-surface meshes from confined 3D point clouds via a physically-based simulation of flexible foils under spatial constraints. The approach integrates dynamic elasticity, pressure-driven deformation, and adaptive snapping to fixed vertices, providing a robust framework for realistic and physically accurate mesh creation. Applications in computer graphics and computational geometry are discussed.