Research On Parallel Collision Detection Based On Delta3D

Collision detection is an important part and supporting technology in the field ofcomputer graphics, virtual reality and3D animation systems. Due to the complexity of virtualscene and the increasing size of the data, the speed and accuracy of collision detection areseriously affected.This article first designs and pulls off a virtual man roaming system basedon the Delta3D game simulation engine, and then studies the parallel collision detectiontechnology and applies it to the virtual man roaming system.The main work of this paper is as follows:Firstly, this paper proposes a optimized24-DOP bounding box model based on SIMD instruction. the model can be accelerated insevaral aspects,including its reconstruction, merging and intersection test. Secondly, ingeometric object vertex updating and bounding box reconstruction, this paper proposes aparallel vertex updating and bounding box reconstruction method based on the OpenMP. Thismethod uses the OpenMP technology to handle geometric object vertex updating andbounding box updating process parallelly, thereby, the calculation speed of collision detectionalgorithm can be improved. Thirdly, this paper presents a parallel collision detectionalgorithm based on general purpose graphics processor unit(GPGPU) on the side ofintersection test, This method turns the intersection test tasks of the basic element over toGPGPU to speed up collision detection. Finally, this paper designs and realizes a virtualhuman roaming system, and extensive experiments are performed to test the frame rate andphysical effects of the system.