Research And Application Of Hybrid Collision Detection Algoriyhm Based On Particle Conversion And Bounding Box

Collision detection technology is an essential part of computer science.This technology is assurance of the fidelity of virtual reality.Therefore,collision detection has become the research focus of computer simulation,virtual reality,robot and other fields.With the rapid improvement of computer technology,the traditional collision detection algorithms are barely meeting the requirements of the efficiency and accuracy.The performance of collision detection algorithms needs to be optimized.In order to improve the efficiency of collision detection,this thesis proposes a hybrid collision detection method which combines particle conversion and traditional bounding box.This method uses both of the AABB and OBB.First,each primitive needs to create the two kinds of bounding boxes separately.AABBs are constructed by axis projection,and OBBs are constructed by using the regional center method.Those bounding boxes are divided by plane segmentation method which divides those objects into layers by splitting planes and axis.A hierarchy bounding box tree is built by putting those divided AABBs and OBBs in a binary tree from top to bottom.Secondly,those bounding boxes need to be updated according to the results of the intersection test which contains six times overlapping test to AABB and an axis test to OBB.OBBs are intersected as long as there is no separated axis in the axis test.As for the hierarchy bounding boxes,a breadth first traverse intersection test is applied to each bounding box.Finally,the bounding boxes are converted to particles by particle conversion formula,and the particles are substitute for bounding boxes in the collision detection step.Those detected particles are reconversion to bounding boxes to apply the intersection test.Four groups of multi-object collision test are experimentalized in order to verify the performance of the improved collision algorithm.Those experiments have different environments including tetrahedron with diverse parameter and diverse scene size.The results of those experiments are used to analyze the time consuming of this algorithm and previous.This thesis also utilities the algorithm in cloth and virtual environment to test the performance.The experimental results show that the algorithm improves the detection efficiency compared with the previous algorithms.With the increasing of scene size,the time consumption in detection will be less consuming,especially a higher efficiency is obtained while there is a large number of objects in the scene.