[postgis-users] TIN from point coverage

Stephen Woodbridge woodbri at swoodbridge.com
Thu Mar 24 16:23:49 PDT 2011

We have a use case for this in pgRouting.

Basically we compute the driving distance of a network and get a 
collection of points that can be represented as x,y,z where z is the 
cost to get to that point.

You can visualize this data like a cereal bowl where the start node is 
in the center of the bowl with a cost of zero and then as you move 
outwards the cost increases as you move up the edges of the bowl.

So we take these points and we need them triangulated.
Then we we intersect the triangles with parallel planes increasing in z 
to get isochronal contour edge pieces, that we could then feed to 
st_BuildArea to create polygons for each isochrone.

But all this starts with Delaunay triangulation.


On 3/24/2011 6:17 PM, Olivier Courtin wrote:
> On Mar 24, 2011, at 10:52 PM, Pierre Racine wrote:
> Pierre,
>> My goal is to define the way leading from a point coverage to a
>> raster. Something like ST_Interpolate("pointgeomtable",
>> "rastercolumn") -> raster. What would be the step leading to that?
>> I guess we have everything now to store a Delaunay Triangulation as
>> polyhedral surfaces right?
> Hum yes and no, in fact it depends :)
> As i said the TIN model is designed for a feature model storage (CIM use
> case),
> so in a single cell, a reasonable small TIN (i.e thousands of points is
> fine
> or even more, but not a whole LIDAR Coverage triangulation)
> If we want to handle this use case, i think
> we have to add a way to split and aggregate a big TIN into several small
> ones
> (as i recall it should be the way Oracle Spatial handle it)
>> Then from this polyhedral surface we can derive a raster...
>> Maybe a first ST_AsDelaunayTriangulation(geom) could be an aggregate
>> SET function returning a set of triangle...
> Yes, if you produce a set of Triangles rather than a single huge TIN,
> it could works with the current TIN implementation.
> Question: is it something liblas could already handle ?
>> and maybe ST_Interpolate(polyhedral surfaces) could be an aggregate
>> function that would aggregate those triangle into a raster?
> --
> Olivier
> _______________________________________________
> postgis-users mailing list
> postgis-users at postgis.refractions.net
> http://postgis.refractions.net/mailman/listinfo/postgis-users

More information about the postgis-users mailing list