| With the development of computer, people have higher requests on the virtual reality technology in immersion, interactivity and real-time. As an important part of virtual reality technology, collision detection technology has also become a hot spot in the study of virtual reality technology. Curved surface is one of the basic graphics in the computer geometric modeling technology. So the collision detection between curved surface in the virtual scene is very common. This paper lucubrates the collision detection technologies which are based on computing the distance between curved surface. Mainly for the following several aspects of research work:1. This paper study the form of expression of non-uniform rational B-spline (NURBS), as well as the meaning of the parameters of it, and confirmed the NURBS surface having the advantage of accuracy and universality in defining general curved surface;2. This paper researched a kind of collision detection technology based on computing the distance between the NURBS surface. At first, it proposed using dynamic splitting decomposition algorithm in splitting NURBS surface, reducing the storage and computation of the non-optimal-solutions. And then, it introduced normal vector to screen redundant nodes in the process of restruction. Finally it determined the pairs of points in the minimum distance between two NURBS surface. Through the experiment, it proved that this method is effective to reduce the node calculation time and improve the efficiency of collision detection and had a good accuracy and robustness;3. The paper also proposes another kind of collision detection technology based on computing the distance between canal surfaces. Using different methods in general canal surface and especial canal surface to compute the distance between them. From the experiment, it confirms this algorithm is feasible;4. Some modern optimization algorithms are summarized in this paper, such as particle swarm algorithm, genetic algorithm and artificial neural network. It proposed using particle swarm optimization algorithm to improve artificial neural network which has the characteristics of parallelism computation, and combined with genetic algorithm in calculating optimum value of a nonlinear function;5. This paper put forward the minimum distance of the control nodes between surfaces could transform into calculating the minimum value of a nonlinear function. Accounting to this method, this paper proposed another collision detection technology which is based on parallel calculating the distance between surfaces. Through the simulation of six pieces of tea fall into cup bottom in the virtual scene, it confirmed this method is feasible, and immersion well. |
