[postgis-users] Producing non-overlapping polygons

[Soeren Meyer-Eppler]

Hello,

I have an application in microscopy (not GIS for once ;-) ) producing huge polygon soups. There may be upwards of 100.000 individual, small polygons. They are mostly disjoint (i.e. their intersection is empty) but since they are produced from lots of overlapping images I can get overlapping polygons at image borders. I want all polygons that touch or overlap to be merged into one. The end result should be a set of non-overlapping, non-touching polygons.

Question: How can I archieve this?

Second question: I'm a noob with geospatial databases. What kind of indices can I use to speed up common queries? 
I tried this:
CREATE INDEX geometryindex ON platelayout USING GIST ( cgeometry GIST_GEOMETRY_OPS ); 
however, a simple geom_union on the table still runs for days (!). I heard about something like cascaded unions - are they implemented yet?
Also what would be the most efficient way to query the polygon soup for all polygons in a user defined, polygon-shaped, region of interest?

best regards,

    Sören Meyer-Eppler

-- 
Cruse Leppelmann  
Kognitionstechnik GmbH      
Technologiehof
Mendelstr. 11 L2
D - 48149 Münster
Web:   http://www.clkgmbh.de
 
Geschäftsführende Gesellschafter:  Dr. Carsten Cruse | Dr. Stefan Leppelmann | Amtsgericht Münster:  HRB 4920 | Umsatzsteuer - ID: DE184124001 | Steuer Nr.: 336/5702/0991