[postgis-users] ST_ConcaveHull giving up too soon, aka too convex

[Rémi Cura]

Hope it works =)

Don't forget to use text paramter of buffers to reduce the number of
segpoints (faster, more coherent with your data)
Try either 'endcap=square'
or quadseg = 1 (I can't test here)
Cheers,

Rémi-C


2014/1/14 Pierre Racine <Pierre.Racine at sbf.ulaval.ca>

> Try this:
>
> SELECT ST_Buffer(ST_Union(ST_Buffer(geom, 0.0001)), -0.00008) FROM points
>
> > -----Original Message-----
> > From: postgis-users-bounces at lists.osgeo.org [mailto:postgis-users-
> > bounces at lists.osgeo.org] On Behalf Of Ivan Price
> > Sent: Tuesday, January 14, 2014 8:36 AM
> > To: PostGIS Users Discussion
> > Subject: Re: [postgis-users] ST_ConcaveHull giving up too soon, aka too
> > convex
> >
> > Thanks for your reply Remi,
> >
> >
> >
> > for now we are going with simply a union of a buffer of the points (as
> you
> > suggested) as it makes sense in the context of the origin of the data
> > anyway.
> >
> >
> >
> > cheers
> >
> >
> >
> > -i
> >
> >
> >
> >
> >
> > De : postgis-users-bounces at lists.osgeo.org [mailto:postgis-users-
> > bounces at lists.osgeo.org] De la part de Rémi Cura
> > Envoyé : Tuesday, 14 January 2014 11:39
> > À : PostGIS Users Discussion
> > Objet : Re: [postgis-users] ST_ConcaveHull giving up too soon, aka too
> > convex
> >
> >
> >
> > Hey,
> >
> > sadly I can't answer your questions,
> >
> > but it looks like you don't want concave hull but instead Alpha Shape
> (see
> > SFCGAL , maybe they have integrated it).
> >
> >
> > You have still a lot of workaround :
> > if you don't mind expending the bouyndary, a union of buffers of points
> > if you don't mind quantization (anyway your input looks quantized) :
> > convert to raster, then any classical image processing easily available
> > (watershed for example) (you can use out of the box QGIS 2.0 raster
> > processing tools)
> > if you want a sharp boudary : and assuming your points are regularly
> > spaced : maybe something liek this (untested)
> > create a segment for each point to his K nearest neighbours (K would be 2
> > or 3, and you would put a max distance before drawing a line)
> >
> > Then pick the segments that have no  intersection with other segments,
> and
> > either use ST_Polygonyze, or cut the bounding box polygon with the
> > segments and remove the part outside
> > .
> >
> > Maybe you can try to reduce your point precision for processing (snap to
> > grid) and/or put the points in an appropriate srid (not too much digits)
> >
> >
> >
> > Cheers,
> >
> > Rémi-C
> >
> >
> >
> > 2014/1/14 Ivan Price <Ivan.Price at noveltis.fr>
> >
> >
> >
> > Hi there,
> >
> >
> >
> > I am experimenting with ST_ConcaveHull, am using Postgres 9.2.2 on
> > windows with postgis 2.1.1 r12113
> >
> >
> >
> > i have a table of 218 points forming roughly an arc (something like the
> nike
> > swoosh)
> >
> >
> >
> > when i ask for the concave hull i'm expecting a poly shrink-wrapped to
> > them, however this is not the case, even when i set a very low percentage
> > area parameter. for example:
> >
> >
> >
> > SELECT ST_ConcaveHull(ST_Collect(geom), 0.000001) as geom FROM
> > points;
> >
> >
> >
> > i attach a screenshot of the points and the convext hull, as well as an
> sql
> > dump of the points themselves.
> >
> >
> >
> > also as a side note, in researching the problem i upgraded from postgis
> 2.01
> > to 2.1.1 and noticed a change in the processing time for the above
> > statement from 1 minute to ~ 5 minutes.
> >
> >
> >
> > are there any pointers for making ConcaveHull more 'aggressive' ?
> >
> >
> >
> > thanks and regards
> >
> >
> >
> > -i
> >
> >
> >
> >
> > _______________________________________________
> > postgis-users mailing list
> > postgis-users at lists.osgeo.org
> > http://lists.osgeo.org/cgi-bin/mailman/listinfo/postgis-users
> >
> >
>
> _______________________________________________
> postgis-users mailing list
> postgis-users at lists.osgeo.org
> http://lists.osgeo.org/cgi-bin/mailman/listinfo/postgis-users
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/postgis-users/attachments/20140114/8a145c11/attachment.html>