[postgis-users] return boundary

[Tim Bowden]

On Mon, 2008-11-24 at 17:15 -0800, Martin Davis wrote:
> They're similar, but I'm not sure they're 100% the same in all cases.
> 
> In the "mainland plus islands" case it seems there is a requirement to 
> ensure that all the input polygons are contained in the result.  The 
> concave hull concept as presented is based purely on a point set, and 
> thus might not necessarily cover all of the input polygons.

Very true, the concave hull may well result in a boundary inside the
original polygon boundary, but then doing a union of the concave hull
and the original polygon would solve this I think (at least for the
bounding polygon.  Not sure what would happen if there are tin's
involved).

> 
> Also, the concave hull is defined by a distance tolerance.  In a 
> situation with islands both very close and very far, it might be hard to 
> find a distance value which would include all islands but not compromise 
> the shape of the main polygon too much.

I suspect it would not be hard to find cases where the optimum approach
(for the current algorithm) would be manual intervention in the form of
breaking the problem (or solutions- easier?) into multiple pieces and
then pick and choosing to stitch them together again to get a final
result.  Awful way to go though.

Regards,
Tim

> Of course it all depends on the data - there's probably lots of cases 
> where both would produce similar results.
> 
> Tim Bowden wrote:
> > On Mon, 2008-11-24 at 11:45 -0500, Travis Kirstine wrote:
> >   
> >> I don't think that the geom union would in combination with the
> >> boundary would attach the island to the mainland.  I don't think that
> >> there is a simple answer to my problem but I have attached a pic to
> >> help illustrate what I'm trying to accomplish.
> >>
> >> Regards
> >>
> >>     
> >
> > Looking at that image seems to me it's just another case of what Regina
> > pointed to (http://ubicomp.algoritmi.uminho.pt/local/concavehull.html)
> > in the concave hull thread.
> >
> > Regards,
> > Tim
> >
> >   
> 
-- 
Experience is that marvelous thing that enables you recognize a mistake
when you make it again.