结构化骨架求取和内窥可见性计算

Structural Volume Skeletonization and Visibility Computation for Virtual Endoscope

提出了包含距离和拓扑信息的结构化骨架提取方法和骨架指导的内窥可见性计算方法,并将其有机地应用于虚拟漫游系统中.结构化骨架提取算法结合了并行细化算法和距离变换算法,使得骨架能够有效地控制虚拟内窥镜的视点移动和漫游位置的跟踪,有利于准确地观察病变位置.内窥可见性算法利用人体器官的封闭管状特征,将其分割成许多网格单元,并在预处理中使用深度缓存计算单元间的可见性,同时对每个单元建立可见性树.虚拟漫游时,通过当前视点信息可进一步动态地删减所在单元的可见性树,从而得到实时可见性,最终实现实时漫游并保持漫游图像的质量.

Two new algorithms of structural volume skeletonization and internal visibility computation are presented to support the virtual endoscope system. The first algorithm extracts the structural skeletons by parallel thinning and distance transform. The second algorithm pre-computes the visibility information of the virtual endoscope to enhance rendering speed. We decompose the elongated human organs into a number of cells; the visibility between cells is then calculated with Z-buffer and stored in a visibility tree for each cell. During navigation, the set of cells potentially visible at the current viewpoint can be easily obtained from the respective visibility tree and further culled according to the precise position of the viewpoint. Finally, realtime navigation with high visual quality can be achieved.