Re: [boost] GSoC2010 Sweepline Algorithm

Subject: Re: [boost] GSoC2010 Sweepline Algorithm
From: Fernando Cacciola (fernando.cacciola_at_[hidden])
Date: 2010-04-12 14:15:16

Hi Andriy,

> So do we need to compute straight skeleton (it consists only of straight
> line segments) or medial axis (may also include parabolic arcs)?
> Medial axis seems to be more similar to Voronoi diagrams (instead of sites
> input it uses edges as Luke said). So I am quite not sure if we should
> compute straight skeleton there.

There is one major thing that is not evident in most drawn samples of a straight 
skeleton: it might very well be anything but "medial" if the polygon is 
non-convex. That is, in a non-convex polygon, the inner bisectors might be 
significantly pushed toward one side of the polygon.

Thus, a medial axis is a subset of a straight skeleton ONLY if the polygon is 
convex. If not, the medial axis is a subset of the voronoi diagram instead.

Best

--
Fernando Cacciola
SciSoft Consulting, Founder
http://www.scisoft-consulting.com