Research Of Collision Detection In Skinned Mesh

Collision detection has been researched in many fields such as robot motion planning and computer graphics for a long time. In recent years, with the rising of virtual reality and distributed interactive simulation, many researches focus on collision detection, and exact collision detection is very important to improve reality and enhance immersion for virtual environment, however, the complexity of virtual environment especially the large scale complex environment brings higher requirement to collision detection.Character animation is among the most active fields in Computer Science in the recent decade, it is widely applied in Virtual Reality, CAD, Advertising and Digital Entertainment, etc. As computer hardware, especially consuming-level video cards with built in acceleration capability technology progresses, Real-time Character Animation is becoming more and more widely used; it is of a great economic prospect and theoretical research significances. Skinned mesh animation is a kind of virtual 3D character rendering technique used extensively recently.It combines the characteristic of single mesh animation and articulated animation and it not only get over the disadvantage of single mesh animation who needs a lot of memory,but also get over the disadvantage of articulated animation who has the problem of crack.This thesis makes a study of collision detection in skinned mesh animation.In previous studies,there weren't spacialized theory about the collision detection of skinned mesh but just using the traditional technology of collision detection to the skinned mesh animation.So they cann't fit the skinned mesh system well.This thesis based on the traditional collision detection technology and using the characteristics of skinned mesh,modified the collsion detection.So can fit the system very well.At the last,validate the applicability of the algorithmic.This paper mainly done the following work:①make use of the structure's feature of bones in skinned mesh animation and choose OBB as the box of virtual character,so the banding box can contact with the bones and fit the bones.②improve the combine algorithmic of two OBB and improve the generation algorithmic of character OBB tree based on it.③make use of the feature of animation struction in skinned mesh animation to improve the update algorithmic of OBB tree.so when update the OBB tree,we can use the information of character's bone to accelerate the update of OBB tree.At the last,we prove that the algorithmic in this thesis can improve the efficiency of collision detection and can satisfy the requirment of collision detection of 3D characters in virtual environment.