Medial hex-meshing: high-quality all-hexahedral mesh generation based on medial mesh

• Published in: Engineering with computers Vol. 40; no. 4; pp. 2537 - 2557
• Main Authors: Zhang, Sheng, Xu, Gang, Wu, Haiyan, Gu, Renshu, Qi, Long, Pang, Yufei
• Format: Journal Article
• Language: English
• Published: London Springer London 01-08-2024; Springer Nature B.V
• Subjects: Approximation; CAE) and Design; Calculus of Variations and Optimal Control; Optimization; Classical Mechanics; Computer Science; Computer-Aided Engineering (CAD; Computers; Control; Math. Applications in Chemistry; Mathematical and Computational Engineering; Mesh generation; Meshing; Methods; Optimization techniques; Original Article; R&D; Research & development; Systems Theory; Cross field; Hexahedral meshing; Medial mesh; Volumetric subdivision
• ISSN: 0177-0667; 1435-5663
• DOI: 10.1007/s00366-023-01925-5

Automatic high-quality all-hexahedral mesh generation is still a challenging problem in engineering applications. In this paper, based on the manifold curve/surface-mixed medial axis representation, we propose a new high-quality all-hex mesh generation method. Given an input watertight model, we first compute the corresponding medial mesh via the medial axis transform simplification method. Then, we build the all-hexahedral layer on the surface skeleton of the medial mesh via cross-field-guided quad-meshing and extrusion, and construct the hex-mesh elements for the curve skeleton of the medial mesh via sweeping approach. Based on the topological and geometrical information of the medial mesh, the initial hexahedral mesh can be obtained. Furthermore, with the iterative volumetric subdivision fitting approach, the hexahedral mesh is fitted to the input model. Finally, padding refinement and mesh optimization method are used to improve the element quality. In order to enhance the robustness and applicability of the proposed method, an interactive framework is also presented to handle non-manifold medial mesh. To show the efficiency of the proposed method, we have extensively tested our method on a lot of models. Compared with existing hexahedral mesh generation methods, our method can generate all-hex meshes with simpler singular structure, better element quality, and smaller element numbers. The code and data will be made available online to foster future research in this field.