Re: [Jts-topo-suite-user] Subtracting lots of small polygons from a large one

[Martin D. <mtn...@gm...>]

The performance improvement for unary union (AKA cascaded union) all
depends on the data it's working on.  For large datasets with a high degree
of overlap it can be *much* faster - up to 50 times or more in extreme
cases.

The way to tell if your data might provide such an improvement is to run
unions in the JTS TestBuilder. It has both possible implementations of
union (cascaded and "brute-force" iterated), so you will be able to see the
difference.

And the Cascaded Union code should not be that hard to port - most of the
hard stuff is done by the existing union and spatial indexing code.

On Fri, Dec 7, 2012 at 2:42 AM, Karel Bruneel <kar...@gm...>wrote:

> Hi Martin,
>
> The holes intersect each other  and the shell, so I'll definitely have to
> go for the second option. The problem there is that  unary union has not
> been fully ported yet in jsts. How much faster do you think it will be with
> the unary union? If its worth it I could dig into the code and try to do
> the port myself.
>
> Thanks
> Karel
>
> On Dec 6, 2012, at 6:35 PM, Martin Davis wrote:
>
> > This is probably obvious, but if you know the holes do not intersect the
> polygon shell or other holes, then you can simply construct a new polygon
> with the holes created explicitly.
> >
> > In the more general case, one way to improve performance is to union all
> the hole polygons first (using Unary Union), and then do a single
> difference operation.
> >
> > On Thu, Dec 6, 2012 at 9:28 AM, Karel Bruneel <kar...@gm...>
> wrote:
> > Hi Everybody,
> > Hi Martin, :)
> >
> > I'm trying to make a polygon with a lot of holes in it and it is
> extremely slow. I'm doing it in jsts (javascript), what makes it even more
> crucial to get the algorithm right.
> >
> > The way I do it now is:
> > - I create the polygon (poly)
> > - and then punch the holes in it one by one using:
> poly.difference(hole_poly)
> >
> > Is there a better way to do this? I've been thinking triangulation, but
> I'm reluctant to try that because I really don't know if it will help.
> >
> > What is the fastest way to punch a lot of holes in a polygon using jts?
> I'm not afraid of some programming as long as there is a decent chance of
> it running fast.
> >
> > Maybe I need an algorithm that is not available in jst at this point?
> >
> > Thanks a lot!
> > Karel
> >
> >
> >
> >
> >
> ------------------------------------------------------------------------------
> > LogMeIn Rescue: Anywhere, Anytime Remote support for IT. Free Trial
> > Remotely access PCs and mobile devices and provide instant support
> > Improve your efficiency, and focus on delivering more value-add services
> > Discover what IT Professionals Know. Rescue delivers
> > http://p.sf.net/sfu/logmein_12329d2d
> > _______________________________________________
> > Jts-topo-suite-user mailing list
> > Jts...@li...
> > https://lists.sourceforge.net/lists/listinfo/jts-topo-suite-user
> >
> >
> ------------------------------------------------------------------------------
> > LogMeIn Rescue: Anywhere, Anytime Remote support for IT. Free Trial
> > Remotely access PCs and mobile devices and provide instant support
> > Improve your efficiency, and focus on delivering more value-add services
> > Discover what IT Professionals Know. Rescue delivers
> >
> http://p.sf.net/sfu/logmein_12329d2d_______________________________________________
> > Jts-topo-suite-user mailing list
> > Jts...@li...
> > https://lists.sourceforge.net/lists/listinfo/jts-topo-suite-user
>
>