Hey Anna Yes, I thought of that but the coarse models I would have access would be the STRM which has 10m vertical accuracy, plus that is Height above ellipsoid I think, so I would have to convert it using the geoid ondulation meaning I would have another error (or am I mixing topics here?). So besides all the difference between data I would have increasing error sources in everything.
Making a DTM based on random points on locations where I know the ground is would lose the ability to detect height variations within high canopy covers but could be an approach since ground heights are otherwise unavailable? All the best! Nuno On 15 May 2015 at 20:45, Anna Petrášová <[email protected]> wrote: > > > On Fri, May 15, 2015 at 10:35 AM, Nuno Sá <[email protected]> wrote: > >> I've got the same questions :) >> >> How will you get the DTM if you don't have info on the ground ? >> >> I did some search for publications but couldn't find if there were any >> methods just by using UAV data, do you have any recommendations? >> >> Ciao! >> Nuno >> >> On 15 May 2015 at 14:40, Adam Laža <[email protected]> wrote: >> >>> Hi all, >>> >>> I've got some data from UAS (SenseFly eBee) exported as point cloud. I'd >>> like to ask if there's any way how to classify the data in GRASS. As an >>> input I have a point cloud exported in .las or .txt file. As an output I >>> need another .las file containing only terrain (I need to filter out >>> objects, buildings and vegetation) for next step which is generating DTM. >>> >>> I've already tried some modules, import(r.in.lidar, v.in.lidar) works >>> well. Then I focused on v.lidar.* modules (v.lidar.edgedetection, >>> v.lidar.growing, v.lidar.correction) but already the first step of >>> recognizing and extracting object didn't work for me. I suppose it's due to >>> the v.lidar.* modules need data only from LiDAR, but I have data from >>> phtogrammetry (eBee carries Canon RGB camera). >>> >>> Any idea how to classify my data? >>> >> > Hi, > > I haven't tried v.lidar.* modules, because I typically work with already > classified lidar, what exactly was the problem? > > Generally, you it would be difficult to get ground from UAV data because > the point cloud represents only surface (unlike lidar where you get > multiple returns). You could compare the point cloud with some coarser > digital terrain model and high differences would mean that there is some > vegetation or buildings. > > Anna > > >>> Thank in advance, >>> Adam >>> >>> Data sample at google drive: >>> las: >>> >>> https://drive.google.com/file/d/0B3qa8r8b0sq0TTdnSVNHdE1UQ2M/view?usp=sharing >>> txt: >>> >>> https://drive.google.com/file/d/0B3qa8r8b0sq0NVFRMmNEZHNkbDA/view?usp=sharing >>> >>> >>> _______________________________________________ >>> grass-user mailing list >>> [email protected] >>> http://lists.osgeo.org/mailman/listinfo/grass-user >>> >> >> >> >> -- >> >> Nuno César de Sá >> +351 91 961 90 37 >> >> >> _______________________________________________ >> grass-user mailing list >> [email protected] >> http://lists.osgeo.org/mailman/listinfo/grass-user >> > > -- Nuno César de Sá +351 91 961 90 37
_______________________________________________ grass-user mailing list [email protected] http://lists.osgeo.org/mailman/listinfo/grass-user
