[postgis-users] How to make this invalid polygon valid one

pcreso at pcreso.com pcreso at pcreso.com
Sat Mar 10 14:02:28 PST 2012


I forget where I learned that... probably from Regina Obe at Boston GIS. 

Also note that from the picture, it should perhaps be a multipolygon, to store a single feature comprising two polygons sharing a point... rather than a single polygon.

Cheers,

  Brent

--- On Sun, 3/11/12, Puneet Kishor <punk.kish at gmail.com> wrote:

From: Puneet Kishor <punk.kish at gmail.com>
Subject: Re: [postgis-users] How to make this invalid polygon valid one
To: "pcreso at pcreso.com" <pcreso at pcreso.com>, "PostGIS Users Discussion" <postgis-users at postgis.refractions.net>
Cc: "PostGIS Users Discussion" <postgis-users at postgis.refractions.net>, "nicolas.ribot at gmail.com" <nicolas.ribot at gmail.com>
Date: Sunday, March 11, 2012, 10:37 AM



ha! Neat trick, Brent.
On Mar 10, 2012, at 3:26 PM, pcreso at pcreso.com wrote:

Hi Nicolas,

Change your SQL to insert a buffered version of your polygon with a buffer distance of zero. You should get a topologically correct polygon (in this particular case anyway), which is a tracing of the perimeter with any duplicate vertices resolved.

eg: 

insert into test_polygon (id,geometry) values (1,ST_buffer(ST_GeomFromText(
'POLYGON((...))'),0.0));

Cheers,

  Brent Wood

--- On Sun, 3/11/12, Nicolas Ribot <nicolas.ribot at gmail.com> wrote:

From: Nicolas Ribot <nicolas.ribot at gmail.com>
Subject: Re:
 [postgis-users] How to make this invalid polygon valid one
To: "PostGIS Users Discussion" <postgis-users at postgis.refractions.net>
Date: Sunday, March 11, 2012, 9:45 AM

Hello



I am trying to create a buffer around my polygon, which unfortunately is not

a valid one.

(reason: "Self-intersection[......]").



This polygon is the output of my calculation that creates from a set of

connected edges an area, consisting of all edges located on the border.



http://postgis.17.n6.nabble.com/file/n4564623/Screen_Shot_2012-03-09_at_10.02.41_PM.png

Polygon image



can someone tell me, what can I do, to remedy this behavior and to make this

geometry valid



this are the sql statements to reproduce my problem:

http://postgis.17.n6.nabble.com/file/n4564623/testPolygon.sql

testPolygon.sql



thanks



Markus




Hi Markus,
An attempt with Pgis 1.5.3:
I saw in a Paul's presentation the power of unioning linestrings together, in order to generated several line where they intersects (http://2010.foss4g.org/presentations_show.php?id=3369)

The goal is to use duplicate points from the initial polygon to allow us to make the difference between these points and the exterior ring of the polygon, to generates several linestrings that can be passed to st_buildArea:

The first part (with...) extracts duplicate points (cf img1)
with duplicates as (      select st_collect((st_dumppoints(geometry)).geom) as point	 from test_polygon 
	 group by (st_dumppoints(geometry)).geom	 having count((st_dumppoints(geometry)).geom)>1
 )  select st_buildarea (      st_union(st_difference(st_exteriorRing(t.geometry), point))   ) from duplicates, test_polygon t;
The result (img2) is a multipolygon that can be dumped to extract individual parts, or directly buffered.
(the query is not at all optimized ;) )
Nicolas

-----Inline Attachment Follows-----

_______________________________________________
postgis-users mailing list
postgis-users at postgis.refractions.net
http://postgis.refractions.net/mailman/listinfo/postgis-users
_______________________________________________
postgis-users mailing list
postgis-users at postgis.refractions.net
http://postgis.refractions.net/mailman/listinfo/postgis-users
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/postgis-users/attachments/20120310/078a3c97/attachment.html>


More information about the postgis-users mailing list