Research On Ray Tracing Using FPGA

Shooting and Bouncing Ray(SBR) algorithm completes the calculation of multiple scattering by shooting ray tracing, the results are more accurate. SBR is widely used in high precise radar cross section(RCS) prediction. Considering the disadvantages such as large amount of calculations, long computing time of the SBR algorithm, this paper firstly uses octree algorithm to optimize SBR, then accelerates the SBR method in the field programmable gate array(FPGA) hardware environment, ultimately,improves the computational efficiency of the algorithm. The research of accelerating SBR algorithm has important applications in the target recognition, system design, stealth and anti-stealth etc.For accelerateing the SBR algorithm based on octree, this paper divides the whole scene to octree according to principle of octree subdivision. After a child node is determined as the tracing object, the shadow area can be identified through the shadow recognization. Then respectively the ray-bounding box, ray-plane intersection tests are computed. Therefore multiple reflection effects can be obtained at a time. Through the comparison and analysis with the classical SBR algorithm, the correctness of the algorithm in this paper is verified and the amount of calculations was simplified significantly.Field tracing of SBR method is based on the shooting ray tracing, to calculate the field strength of every reflection ray. The final field strength is computed thougth accumulating the emergent field of all rays. Eventually, this paper calculates the target RCS prediction combining with physical optics and far field integral calculation.In order to accelerate the SBR further, this paper proposes a SBR accelerating method based on the FPGA platform. A Static Random Access Memory(SRAM) is employed for memorying the whole scene information including all points coordinates and all triangles. The designs of ray tracing, the field integral calculation module based on FPGA, can be real-time access to the SRAM. Comparing to the Central Processing Unit(CPU), this method is more effective. Finally comparing the simulation results based on the CPU, the hardware simulation results is not only high-precision, but also greatly improves the computational efficiency. It can reach about 600 times speedups than the traditional sequential algorithm. The shooting and bouncing ray method proposed in this paper can ensure real-time in a certain extent but also expand the scope of application.