[OSGeo-Discuss] open source polygon cluster aggregation algorithm?
G. Allegri
giohappy at gmail.com
Thu May 6 05:01:48 PDT 2010
<Aside>
A nice implementation of alpha shapes with jts:
http://www.mail-archive.com/jts-devel@lists.jump-project.org/msg01019.html
</Aside>
2010/5/6 G. Allegri <giohappy at gmail.com>:
> Thanks Andrea for the links. Yes, I think the problem is similar, in
> fact I was also looking for concave hull and alpha shapes algoithms,
> but the only open solution I've found is from CGAL [1] and... it's too
> complex to extract and reimplement in my context (database procedural
> programming).
>
> IWe have implemented something very "rude":
>
> 1 - logically aggregate polyongs in clusters (given a certain distance)
> 2 - buffer each polygon mantaining the shape (not the usual buffer,
> which make rounded artifacts)
> 3 - geometrical union
> 4 - shrink the result (unbuffer)
>
> But I have the time for a long holiday waiting the end of the process :)
>
> [1] http://www.cgal.org/Manual/last/doc_html/cgal_manual/Alpha_shapes_2/Chapter_main.html
>
> 2010/5/6 Andrea Aime <aaime at opengeo.org>:
>> G. Allegri ha scritto:
>>>
>>> I'm looking for an algorithm to do polygon cluster aggregation,
>>> similar to the ArcInfo "Aggregate Polygon" [1].
>>> I know about GEOS "Cascaded Union", but I need two more features:
>>>
>>> 1 - clustering of polygons that fall within a a certain threshold
>>> distance from each other
>>> 2 - mantain orthogonality, i.e. the original angles/shapes
>>
>> I don't know of any such implementation, but it looks somewhat
>> similar to the computation of a concave hull:
>>
>> http://ubicomp.algoritmi.uminho.pt/local/concavehull.html
>> http://stackoverflow.com/questions/83593/is-there-an-efficient-algorithm-to-generate-a-2d-concave-hull
>>
>> Cheers
>> Andrea
>>
>> --
>> Andrea Aime
>> OpenGeo - http://opengeo.org
>> Expert service straight from the developers.
>> _______________________________________________
>> Discuss mailing list
>> Discuss at lists.osgeo.org
>> http://lists.osgeo.org/mailman/listinfo/discuss
>>
>
More information about the Discuss
mailing list