In this paper we give upper and lower bounds on the number of Steiner points required to construct a strictly convex quadrilateral mesh for a planar point set. In particular, we show that 3 n/2 internal Steiner points are always sufficient for a convex quadrilateral mesh of n points in the plane. Furthermore, for any given n ≥ 4, there are point sets for which (n − 3)/2 − 1 Steiner points are necessary for a convex quadrilateral mesh.
Introduction.Discrete approximations of a surface or volume are necessary in numerous applications. Some examples are models of human organs in medical imaging, terrain models in GIS, or models of parts in a CAD/CAM system. These applications typically assume that the geometric domain under consideration is divided into small, simple pieces called finite elements. The collection of finite elements is referred to as a mesh. For several applications, quadrilateral/hexahedral mesh elements are preferred over triangles/tetrahedra owing to their numerous benefits, both geometric and numerical; for example, quadrilateral meshes give lower approximation errors in finite element methods for elasticity analysis [1], [3] or metal forming processes [15]. However, much less is known about quadrilateralizations and hexahedralizations and, in general, highquality quadrilateral/hexahedral (quad/hex) meshes are harder to generate than good triangular/tetrahedral (tri/tet) ones. Indeed, there are several important open questions, both combinatorial as well as algorithmic, about quad/hex meshes for sets of objects such as polygons, points, etc., even in two dimensions. Whereas triangulations of polygons and two-dimensional point sets and tetrahedralizations of three-dimensional point sets and convex polyhedra always exist (not so for non-convex polyhedra [29]), quadrilateralizations of two-dimensional point sets do not. Hence it becomes necessary to add extra points, called Steiner points, to the geometric domain. This raises the issue of bound-