This paper presents an accurate and efficient method for the computation of both point projection and inversion onto Bézier surfaces. First, these two problems are formulated in terms of solution of a polynomial equation with u and v variables expressed in the Bernstein basis. Then, based on subdivision of the Bézier surface and the recursive quadtree decomposition, a novel solution method is proposed. The computation of point projection is shown to be equivalent to the geometrically intuitive intersection of a surface with the u-v plane. Finally, by comparing the distances between the test point and the candidate points, the closest point is found. Examples illustrate the feasibility of this method.Keywords point projection, point inversion, Bézier surface
½ ÁÒØÖÓ Ù Ø ÓÒProjecting a test point onto a parametric surface in order to find the closest point (the point projection problem) and computing the corresponding parameter values of the projection (the point inversion problem ) are two basic problems in computational geometry, geometric modeling, computer graphics, and related topics. Both projection and inversion are useful for surface intersection [1], tool path generation, and collision detection in numerical control (NC) machining [2]. They are also a key issue in the inspection of manufactured objects [3]. For these purposes, it is important to have a computational method which is efficient and reliable to find Several algorithms which employ some variations of Newton iteration or numerical optimization have been developed to solve this problem. Mortenson [4] derived equations for different types of surface distance measure and employed the Newton-Raphson method to solve them. Limaiem and Trochu [5] proposed another approach to compute the projection of a point onto parametric curves and surfaces by constructing an auxiliary function and finding its zeros. These algorithms converge quickly, but the sensitivity of the iteration process to initial values is also well known. As an improvement, Hu and Wallner [6] developed a second-order algorithm for point projection onto curves and surfaces. It increases the robustness to the choice of initial values.This Newton-type methods have been used in CAD/CAM applications and acquired high accuracy because they are efficient (usually exhibiting quadratic convergence rates close to simple roots) and straightforward to program. Unfortunately, this is an error-prone process which often fails for points near the endpoints of the curve or the boundaries of the surface and this procedure typically requires a good initial value to ensure the convergence to the optimal root in the global scene. Such initial approximations are usually obtained by sampling the curve or surface, but this process cannot provide full assurance that all roots have been found. This lack of robustness promotes the development of efficient and reliable techniques. Piegl and Tiller [7] presented an algorithm for point projection on non-uniform rational Bsplines (NURBS) surface. It consists...