Expanding the Solvable Space of Polycube-Map Via Validity-Enhanced Construction
Lu He,Na Lei,Ziliang Wang,Chen Wang,Xiaopeng Zheng,Zhongxuan Luo
DOI: https://doi.org/10.1137/1.9781611978001.4
2024-01-01
Abstract:Polycube-maps are highly valuable in computer graphics, particularly concerning hexahedral meshes. Currently, the validity of polycube is determined based on Steinitz and Eppstein's approach, which only addresses 3-connected graphs. However, extending the features of polycubes to encompass 4- and 6-connected graphs holds significant practical importance. We strengthen the validity structure of the polycube-map, providing a polycube validity theorem based on the Gauss-Bonnet theorem, offering both global and local conditions. Our theorem allows for 4- and 6-connected cases, expanding the existing solvable space of 3- and 5-singularity polycube-map to include 3-, 5-, and 6-singularity scenarios. Our theorem effectively addresses models that, while not meeting the existing valid polycube criteria, are indeed valid polycube polyhedra in practice. Additionally, within the label optimization framework, we introduce the Immune Genetic Algorithm (I-GA algorithm) tailored to our theorem, enabling the robust generation of polycube-map. We evaluate our method using thingi10k and ABC datasets. Results demonstrate that our validity theorem expands the solvable space of polycubes while achieving higher quality all-hexahedral meshing for models with a singularity level of six. Furthermore, we discuss the limitations associated with our proposed method.