Topologically sweeping an arrangement

Herbert Edelsbrunner, Leonidas Guibas

Journal of Computer and System Sciences · 1989 · 171 citations · 15 references

Concepts

TL;DR

Sweeping collections of figures with a straight line is a novel computational‑geometry paradigm, and the authors’ technique, while planar, also applies to higher‑dimensional problems. The paper demonstrates the benefits of sweeping with a topological (non‑straight) line. The authors sweep an arrangement of n lines in O(n²) time and O(n) space, visiting each vertex, edge, or region once in a consistent order, and apply the method via duality transforms to problems such as visibility, degeneracy detection, and computing extremal shadows of convex polytopes. Their algorithm, using novel data structures with favorable amortized complexity, outperforms previous methods in either time or space while remaining practical.

Abstract

Abstract Sweeping a collection of figures in the Euclidean plane with a straight line is one of the novel algorithmic paradigms that have emerged in the field of computational geometry. In this paper we demonstrate the advantages of sweeping with a topological line that is not necessarily straight. We show how an arrangement of n lines in the plane can be swept over in O ( n 2 ) time and O(n) space by a such a line. In the process each element, i.e., vertex, edge, or region, is visited once in a consistent ordering. Our technique makes use of novel data structures which exhibit interesting amortized complexity behavior; the result is an algorithm that improves upon all its predecessors either in the space or the time bounds, as well as being eminently practical. Numerous applications of the technique to problems in computational geometry are given—many through the use of duality transforms. Examples include solving visibility problems, detecting degeneracies in configurations, computing the extremal shadows of convex polytopes, and others. Even though our basic technique solves a planar problem, its applications include several problems in higher dimensions.

References

15