A sah-kd tree design method for ray tracing

Abstract

The present invention provides a SAH-KD tree design method for ray tracing, comprising: step S1, before performing the construction of the SAH-KD tree, preprocessing all triangle faces of the current root node in the node ; Step S2, recursively constructing the SAH-KD tree hierarchically, generating an ordered bounding box sequence of the left and right child nodes of the current node, and calculating three dimensions of the bounding box of each node at the current level surface area; step S3, find out the dimensions that the current node needs to divide, generate the space tree and candidate division points of all the nodes in each layer, and calculate the SAH value corresponding to each candidate division point; step S4, select the minimum The candidate division point corresponding to the SAH value is used as the optimal division point of the current node, and the current node is divided into left and right child nodes. Compared with related technologies, the SAH-KD tree constructed by the SAH-KD tree design method for ray tracing provided by the present invention has good performance and high efficiency.

Description

technical field [0001] The invention relates to the technical field of data processing, in particular to a SAH-KD tree design method for ray tracing. Background technique [0002] K-dimensional tree (KD tree) is a kind of balanced binary tree aligned with coordinate axes. It is a data structure commonly used for space division in K-dimensional space. It is widely used as an acceleration structure in graphics applications and query fields, such as ray tracing. Intersection detection between triangular faces and rays, nearest neighbor query, etc. The traversal cost of the KD tree has different calculation methods in different application fields. In ray tracing, it is necessary to calculate all the leaf nodes of the KD tree that each ray passes through. In order to achieve efficient traversal, the surface area heuristic partition method (SAH) is usually used during construction, and the partition point is selected by minimizing the intersection test complexity between the tria...

AI Technical Summary

Problems solved by technology

[0003] In related technologies, in order to quickly and efficiently construct the SAH-KD tree used for ray tracing, one method is a serial construction method, and the simplest idea is to fully perform all division points on each node of the constructed SAH-KD tree Traversing to calculate the SAH value, its computational complexity is O(N 2 ), due to the high computational complexity, this method cannot be applied to large scenes; the other is the parallel construction method, which uses a high-performance thread to perform calculations on the upper-level large nodes until the calculation amount of each node is small enough, then the They are assigned to the construction idea of ​​the multi-core system. Although this method improves the construction efficiency of the SAH-KD tree, the efficiency of the median selection at the top layer is somewhat degraded when the algorithm is traversed, and there are different degrees of performance loss in super-large scenarios.

Images