医学体数据中基于局部特征尺寸的Delaunay四面体化算法*

为了从医学体数据构建面向虚拟手术仿真系统的器官实体模型，提出一种基于局部特征尺寸的Delaunay四面体化算法。首先采用Marching Cubes算法和外存模型简化技术从体数据中得到器官等值面简化模型，提出重心射线法去除内部冗余网格，获得器官多面体表面；然后基于局部特征尺寸构建表面顶点保护球，结合Delaunay细分算法生成边界一致的初始四面体网格；最后提出基于随机扰动的空间分解法快速生成内部节点，并逐点插入到四面体网格中优化单元质量。该算法克服了Delaunay细分算法无法处理锐角输入的缺点，并从理论

Local feature size based Delaunay tetrahedral mesh generation from medical volume data

This paper presented a Delaunay algorithm based on local feature size to generate tetrahedral mesh from medical volume data for virtual surgery system. Firstly, preprocessed the medical volume data to create polyhedron representation of the domain, including iso-surface extraction, mesh simplification and internal redundant mesh deletion. Secondly,constructed local feature size based protecting balls for all vertices of the polyhedron. Conforming initial tetrahedral mesh was then generated by revised Delaunay refinement method. Lastly, mesh optimization combining space-disassembling method and random perturbations was applied to place internal points quickly and efficiently. It inserted these points one by one to improve the mesh quality. Allowed arbitrarily small input angles in the presented method and the convergence could be guaranteed by theoretical proof. Experimental results show that the presented algorithm has better boundary consistency than the traditional delaunay algorithm. And the generated meshes are well suited for virtual surgery system.