Study of Shooting and Bouncing Ray Method Based on BVH Tree and GPU Acceleration

To improve the computational efficiency of the shooting and bouncing ray (SBR) method, a graphics processing unit (GPU) parallel accelerated computation using the C++AMP platform was implemented in this paper. Using Morton code to convert the triangle data into a one-dimensional linear space maintains the spatial proximity of each triangle. This optimization enhances the efficiency of spatial retrieval and data querying, as well as improves the efficiency of constructing bounding volume hierarchies (BVH) trees and collision detection. By comparing with the ray launching geometrical optics (RL-GO) method, this paper to enhance the computational efficiency of the SBR method, and the radar cross section (RCS) results are in agreement with RL-GO. Inverse synthetic aperture radar (ISAR) images of an aircraft target were calculated, and the imaging results demonstrate that this paper can correctly reflect the target's structural information, highlighting the effectiveness of the method in this paper.