Abstract
The paper provides a set of algorithms for performing set operations on polygonal regions in the plane using standard floating-point arithmetic. The algorithms are
robust, guaranteeing both topological consistency and numerical accuracy. Each polygon edge is modelled as an implicit or explicit polygonal curve which stays within some distance ß of the original line segment. If the curve is implicit, ß is bounded by a small multiple of the rounding unit. If the curves are explicit, the bound on ß may grow with the number of curves. One can mix implicit and explicit representations to suit the application.
