一种用于光线与三角形网格求交运算的有效剔除算法(英文)

An Efficient Rejection Test for Ray/Triangle Mesh Intersection

提出一种用于光线与三角形网格求交运算中的有效剔除算法.算法中,一根光线被定义为两个非平行平面的交线.针对由稠密三角形网格组成的复杂场景,算法通过三角形和测试平面的相交判断剔除与投射光线不相交的绝大多数三角面片.利用该算法,光线跟踪中主光线在图像空间的相关性可以方便、直观地被利用.为了利用物体在景物空间的相关性,算法可以结合层次包围盒、八叉树等常见的场景划分方法.而且,该算法可以方便地扩展应用于一般多边形网格.

This paper proposes a new rejection test for accelerating ray/triangle mesh intersection. In the approach, a ray is defined as the intersection of two nonparallel planes. For a given ray and a complex scene including dense triangle meshes, this approach can cull most nonintersecting triangles by a simple rejection test that only involves triangle/plane intersection tests. With this approach, exploiting image-space coherences for primary rays in ray tracing is straightforward. In order to exploit object-space coherences, the approach can also be combined with popular spatial partition schemes, e.g. bounding box hierarchies and octrees. Furthermore, this approach can be easily extended to more general polygonal meshes.