Shrink wrap meshing via graph-based optimization

In numerical simulations, creating a watertight surface mesh is crucial for subsequent volumetric discretization and numerical analysis. Traditional workflows face significant challenges when processing Computer-Aided Design (CAD) models containing gaps or intersections, typically requiring extensive manual repair efforts prior to meshing operations. This paper presents an innovative feature-aware surface meshing framework that directly processes raw CAD input while eliminating the need for preliminary geometry repair. It represents the mesh as a graph and optimizes vertex positions through a specifically designed hybrid loss function, enabling automatic mesh wrapping through vertex position movement. By integrating mesh projection and topological stitching, the algorithm effectively maintains critical geometric features throughout the optimization process. Our framework prioritizes three essential aspects: guaranteed mesh quality metrics, robust defect tolerance, and adaptive feature preservation. Comprehensive experimental evaluations demonstrate that the proposed method generates meshes with improved element quality while exhibiting superior robustness against geometric defects compared to current state-of-the-art approaches. This advancement provides computational engineers with a promising and automatic surface meshing solution for raw CAD models.