[GRASS-user] Using v.lidar.* for object recognition

[LeeDaniel]

Hi all,

As far as I understand, the v.lidar* modules help to separate surface
objects out from the terrain. I've got a lidar point cloud that I would like
to use to detect building footprints and can't seem to get them to work.

I've followed the instructions on the wiki and cleaned the dataset (there
were basically no outliers) and then used, in sequence,
v.lidar.edgedetection, v.lidar.growing and v.lidar.correction to try to
separate the buildings from the terrain. I was especially excited because I
was hoping to simultaneously be able to differentiate between trees and
buildings, something that II haven't been able to reliably do using only the
DTM and DSM. However, the outputs aren't any different from the inputs,
except that the attribute tables only contain two fields, which contradicts
the help texts for the individual tools.

As far as I can tell from the help texts, it seems that after running the
three tools I should get classified points back that have four categories,
terrain single pulse, terrain double pulse, and object single and double
pulse. Every single point in my results has the category 1, which I don't
understand because if I interpolate from the objects that were preclassified
as non-terrain objects, I can see that there are differences to the
interpolated results if I make an elevation raster from the preclassified
terrain points.

Also, I'm kind of unsure as to how GRASS should know what points were double
and which ones were single pulse.

Could anybody help me further, or does anyone have ideas as to how to to
extract building footprints without extracting the tree footprints too?
Thanks a bunch!

Daniel

--
View this message in context: http://osgeo-org.1803224.n2.nabble.com/Using-v-lidar-for-object-recognition-tp6411300p6411300.html
Sent from the Grass - Users mailing list archive at Nabble.com.