[GRASS-user] From pointcloud to dem - help

[Hamish]

Jorn Anke:
> > I have a pointcloud form PMVS
> > (http://grail.cs.washington.edu/software/pmvs/)
> > consisting of near to 6 million points.

see http://grass.osgeo.org/wiki/LIDAR

specifically the Snake River micro-tutorial section about DEM
creation (but ignore the irrelevant LAS stuff there).

> > The pointclouddata is georeferenced into world-coordinates
> > (EUREF-89), and stored in a text file with the following
> > format: X Y Z R G B. (RGB colordata
> > can be deleted if required).
> >
> > What I want to do is to import the pointcloud into
> > Grass 7.0, and to generate a surface modell/ dem, and to
> > export the dem to a common file-format, e.g. ESRI SHP or
> > Land-XML.

unless you have a real need for it, I'd suggest to stick to the
stable GRASS 6.4 release.

> > I am quite new to Grass,

In that case I'd strongly recommend you stay with a 6.4 release.

> > so I would very much appreciate a brief description
> > of the steps I will need to go through in order to
> > obtain this.

Markus Metz wrote:
> First, import the point cloud with v.in.ascii.

that will choke for more than ~ 3 million points unless you start
without topology, or have several GB of RAM and a bit of time.
Importing without topology is a trick I'd reserve for users
already well familiar with the GIS, as lots of things don't work
right if topology is missing.

> You can then filter the points with the lidar tools to remove
> outliers (v.outlier) and obtain a digital terrain model and a
> digital surface model (v.lidar.edgedetection, v.lidar.growing,
> v.lidar.correction). The v.surf.* modules can create a
> (raster) DEM by interpolating the points.

In this case advanced filtering doesn't seem necessary, so I'd
recommend import directly to raster with r.in.xyz, which can
handle an arbitrary number of input points (hundreds of billions
anyway). First read up on the "raster intro" quick start document
as a guide on how to set up your raster grid bounds and
resolution to suit ("region" in grass parlance).


regards,
Hamish