Elsevier

Computers & Graphics

Volume 26, Issue 1, February 2002, Pages 67-74
Computers & Graphics

Technical Section
Termination criterion for subdivision of triangular Bézier patch

https://doi.org/10.1016/S0097-8493(01)00158-3Get rights and content

Abstract

A theorem regarding the termination criterion for subdivision of triangular Bézier patch is presented with detailed proof. As the number of times for subdividing a triangular Bézier patch can be a determined prior, according to the given tolerance, the subdividing process can be conduced consecutively without checking for linearity after each subdivision. The theorem has potentials in various CAD/CAM applications.

Introduction

Triangular Bézier patch was first introduced to CAGD by de Casteljau in internal technical reports (de Casteljau, 1959–1963). Since then, a great deal of work has been reported in these fields. Sabin [1] studied the triangular Bézier patch independently and his representation scheme is well known.

Subdivision of curves and surfaces are important operations in CAD/CAM because they produce sequences of linear segments or planar facets approximating the curves and surfaces within a given tolerance. Lane and Riesenfeld [2] proposed a termination criterion for subdivision of the Bézier curve and Wang [3] addressed the convergence problem regarding piecewise linear approximation of rational Bézier curve. Goldman [4] studied subdivision algorithms for triangular Bézier patch in 1983 and Chang, Davis [5] gave a practical formula for calculating the control vertices of the triangular sub-patches in 1984.

One interesting problem for subdivision operations is how to determine the number of subdivision times of a triangular Bézier patch a prior based on a given tolerance. In this paper, we present a theorem (Theorem 3) to deduce such a termination criterion.

Section snippets

Preliminaries

Assume a barycentric coordinate system over a domain triangle T:=ΔT1T2T3 such that T1=(1,0,0), T2=(0,1,0) and T3=(0,0,1), a triangular Bézier patch Bn(u,v,w) of degree n over domain triangle T can be defined asBn(u,v,w)=0⩽i,j,k⩽n,i+j+k=nBijkn(u,v,w)fijk(0⩽u,v,w⩽1,u+v+w=1),where fijk(i+j+k=n) denotes the control vertices of the triangular Bézier patch whose number amounts to (n+1)(n+2)/2 and Bijkn(u,v,w)=n!/i!j!k!uivjwk are the respective Bernstein basic functions.

Let D(r,s,t) be the 3D

Estimating the deviation of a triangular Bézier patch from its base triangle

Theorem 3

Suppose that Bn(u,v,w) be a triangular Bézier patch which is defined over domain triangle T(TT1T2T3) and ΔP1P2P3 be a triangle inside T with its edges parallel to that of ΔT1T2T3 (see Fig. 1, Fig. 2 and the barycentric coordinates of its vertices be (ui,vi,wi), i=1,2,3. Let Bn(u*,v*,w*) be a sub-patch of Bn(u,v,w) corresponding to ΔP1P2P3, then the maximum normal distance of Bn(u*,v*,w*) to its base triangle D(r*,s*,t*) will be no larger than a given tolerance ε, i.e.,

d(Bn(u*,v*,w*),D(r*,s*,t*

Geometric interpretations of Theorem 3

Assume that a triangular Bézier patch is subdivided into r2(rr0)sub-patches, then we can use 3D triangular facets defined by the corner vertices of these sub-patches to approximate the original triangular Bézier patch within the given tolerance ε. Therefore, Theorem 3 can be applied to intersection calculation, FEM mesh generation and STL output for triangular Bézier patch-based CAD systems.

An example is shown in Fig. 3. In this figure, a quartic triangular Bézier patch approximating a 18

Acknowledgements

The work is jointly supported by the National Advanced Technology Project of China (No. 863-511-942-018) and the Research Fund for the Doctoral Program of Higher Education (No. 98033532).

References (6)

There are more references available in the full text version of this article.

Cited by (6)

View full text