Transformation Semantics: An Efficient Approach for Collision Detection

Abstract

Collision detection is an important problem in many kinds of applications. This work presents an efficient approach for exact collision detection between complex, deformable models. The approach consists of a method for fast extraction of semantics from transformation matrices which places models in a scene, together with a general strategy for fast intersection tests. Non-uniform scaling is also supported efficiently. Our experiments demonstrate that our strategies are well suitable for real-time applications