Stackless KD-Tree Traversal For Ray Tracing

In ray tracing, the computation takes a long time due to the large number of rays and the complex target. In order to avoid invalid intersection calculation and speed up the intersection test between ray and target scene, the geometric information in space scene can be arranged by establishing KD-Tree (K-Dimensional tree) to speed up the intersection process. This paper proposes a stackless KD-Tree traversal based on ropes. By connecting the nodes of KD-Tree according to certain rules through “ropes”, a high-quality KD-Tree is constructed. In the process of ray tracing, the established ropes can be used to judge the KD-Tree nodes through which the rays pass. The results show that this method reduces the number of searching layers of KD-Tree compared with traditional methods.