基于点索引的网格模型的层次结构

针对稠密采样的网格模型．提出一种基于面片中心点索引的新的场景BSP树结构．与常规RSP树构造方式不同，文中RSP树以面片中心点位置作为场景中各面片二叉分类的依据，避免了常规BSP树构造方法中因分割与削分平面相交的面片引起的场景复杂度的增加，大大简化了BSP树的构造过程．实验结果表明：对稠密网格场景，文中的BSP树比常规方式构造的BSP树任加速光线跟踪算法中的求交测试具有明显的优势．

A New BSP Model for Large Scale Meshes with Point Index

Oriented to the scene composed of densely sampled meshes, this paper proposes a new scheme for binary space partition, which provides only references to the facets that occupy each partitioned space. Rather than sorting the facets of meshes according to their relative positions against the partitioning plane during the recursive space partitioning, as performed by all the current BSP algorithms, our scheme sorts the facets according to the relative location of the center point of each facet, and adopts these points for indexing. Since no facet needs to be split into small fragments, our partition scheme not only reduces the complexity of the scene BSP tree, but also yields significant time savings in BSP tree construction. Furthermore, as illustrated by our initial experimental results, the rendering time for ray-tracing a complex scene with our BSP tree is also reduced in most cases.