Mesh Feature Recovery System Based On Uniform Grid

In many applications of graphics, CAD and FEA, with its specific geometric characteristics, feature is always an essential part of geometry model and is widely used in geometry modeling, mesh segmentation, mesh deformation and geometry processing control. The discretization of the geometrical model and point cloud reconstruction are two main mesh generation methods, but due to the low sampling and bad tessellation, the feature is usually lost, which not only cannot represent the model well but also has an effect on further geometry processing. Therefore, it is important to recover the feature of geometry models.To recover the feature, we propose a feature recovery method based on uniform grid, which can robustly process large-scale mesh model. A mesh model without noise and a series of extra features are expected as input in our method. The basic idea of our method is that we will find a one-one mapping relationship between extra features and surface mesh through projecting features into surface mesh, with this relationship we can recover the feature of mesh.In order to keep the one-one mapping relationship, we analyze the complexity of feature projection and propose some methods based on half-edge data structure and unit operation to deal with the conflict and ambiguity. Moreover, based on the local characteristic of our method we use a uniform space partition data structure to deal with large-scale mesh model. Finally, we develop a robust feature recovery system for large-scale mesh.