[SoC] Distance Analysis for PostGIS Raster - Report Week 4 (Qing Liu)

[Pierre Racine]

Qing,

I find the constraints are not really well stated as a first part of the document. Could you list those constraints, right here, so we agree on them?

Also: 

To produce a distance raster we need to know which point is the nearest from the pixel centroid.

To produce a more generic interpolation surface (as we would like to be able to do), which points do we need to know? Only the three nearest points? The three nearest points forming a triangle encompassing the pixel centroid? More points?

Maybe those two problems can resume to one if we would build a Delauney triangulation with the point before trying to produce any surface. So the process would be two steps: 1) Build a Delauney surface 2) Compute some metric for each pixel centroid.

Mir, Nouri, does this make sense?

Qing, also please port the proposal to the wiki so we can add comments directly in it and from now on we will try to include Mir and Nouri in our discussions. The postgis-devel list should be for more general discussion.

Pierre

> -----Original Message-----
> From: Qing Liu [mailto:liu.qing.1984 at gmail.com]
> Sent: Saturday, June 16, 2012 11:41 AM
> To: soc at lists.osgeo.org
> Cc: Pierre Racine
> Subject: Distance Analysis for PostGIS Raster - Report Week 4 (Qing Liu)
> 
> Hi,
> 
> Here's my Report for Week 3
> 
> http://trac.osgeo.org/postgis/wiki/PostGIS_Raster_SoC_Idea_2012/Distance_A
> nalysis_Tools#Week4Report
> 
> Have a good weekend,
> Qing