[Qgis-user] Any ideas? Splitting polygons to only have one point in each polygon

[magerlin]

Hi - I've tried the one building a time approach but have not been able to
get through.
It is very kind of you Victor to offer to have a look at it and perhaps
include it in your blog.

You can get my data and model files here:  Get data
<https://dl.dropbox.com/u/8189384/Break%20Polygons%20by%20Points.zip>  
The zip file includes both some test data (looking almost like the picture
in my first post) but also some nasty real data with about 14.000 buildings!
Also two model files are included: The one used to create the picture in my
first post and my attempt to the iterative procedure.

Executing the "Break Polygons by Points iteratively.model" give me an error
message:

<http://osgeo-org.1560.n6.nabble.com/file/n5034373/Error_msg.png> . 
There are no messages in the log.

My general experiences so far has been:
* I need to use the Grass v.voronoi since the Qgis "Voronoi polygons" uses
the whole layer even if selections exist
* v.voronoi stops with an error message if less than 3 points in the
building but it has problems handling situations also with bigger (but
small) numbers. In the example below with 5 points it has only made 3
voronoi polygons:
<http://osgeo-org.1560.n6.nabble.com/file/n5034373/5_points_not_ok.png>  
Some other situations with 5 points are handled correctly!
* Due to the above mentioned limitations and since I need to be able to also
split buildings with only 2 points I have experienced that the addition of
some temporary additional points at the (far) edge of the Analysis Extent
will help when they are included in the creation of Veronoi polygons for
every building
* With the large number of polygons in my real data I think the iterative
process at each step should append the new Voronoi polygons inside the
building to a "collecting" layer of split buildings (at the end beeing the
result layer) - otherwise the algorithm must at the end be able to merge an
unknown big number of shape files.


My experiences with Sextante: It is marvellous to now be able to make your
own routines and it is quite simple with the modeler! I have no knowledge of
python so I have not come to the scripts yet.

I have an undocumented feeling that Sextante in the background "saves" some
setting and sometimes unintentionally (at least from my point of view ;-)
keeps them for later runs of other geoalgorithms? 
Sometimes it could seem to be due to (not reported) problems with
overwriting existing files. I have experienced several times that my
"Break_Polygons_by_Points.model" stops to function (nothing happens when
running it) but closing down Qgis and restarting makes it OK again. It could
perhaps also be some saved analysis extents?







-----
Regards Morten

Qgis 1.8.0 Stand alone installer in Win7 64 bit
--
View this message in context: http://osgeo-org.1560.n6.nabble.com/Any-ideas-Splitting-polygons-to-only-have-one-point-in-each-polygon-tp5034081p5034373.html
Sent from the Quantum GIS - User mailing list archive at Nabble.com.