关键词: 碰撞检测, 蜂窝状空间分解, 反向投影, 阈值线

Abstract: Real-time and accurate collision detection research has become a hot research problem in the field of virtual reality.In this paper，an improved honeycomb-shape spatial decomposition collision detection algorithm based on image space is proposed.The non-collision polyhedron is first removed and the collision polyhedron is partially and spatially decomposed.Then back-projection is used to build local sub-collision bounding box and to complete the honeycomb-shape space decomposition.So the collision detection is converted to measure the distance from the cross-section in the sub-cellular space polyhedron of next frame to the threshold line of the other party.This algorithm gets rid of the constraints of graphics hardware，and makes the detection algorithm based on image space fit for arbitrary polyhedra.It improves the speed and accuracy of collision detection.

Key words: collision detection, honeycomb-shape spatial decomposition, back-projection, threshold line