[Liblas-devel] Re: liblas svn and gdal/ogr

[Jeff Hamann]

My understanding is that lidar data *is* point data that is processed  
into something else like raster files, information, "data" (DTMs, etc.).

I would say that the number of points may be so high that it looks  
like continuos data, but in fact, it's point data and the high volume  
of points covering an area might be considered raster data for  
generating surfaces. But what if the surface is very rough (trees),  
and the objective is to find local heights, and not create a smooth  
surface.

I've been building an experimental software agent, using open source  
software, that ingests data and uses the results for optimization and  
I need process at least 1TB of new lidar data annually.

The process right now is as follows:

1) convert a list of LAS files into shapefiles
2) convert the shapefiles to sql for using within PostgreSQL+PostGIS 
+wktraster+PL/R+plus a ton of other stuff
3) import the sql file into the database.

This process takes several hours for a single file and requires that I  
duplicate the data twice (at least).

Since my shop is just me, I need to generate a ogr2ogr type tool that  
can be used like:

$ ogr2ogr -f "PostgreSQL" PG:"host=localhost dbname=himom"  
'biggen.las' -nln test

so that the lidar data goes directly into the database.

I would love to hack on this myself, and contribute all of the  
projects which I currently try to rely upon, but cash flow is  
negative, and has been for 24 months, so spending a few weeks hacking  
on open source projects just isn't in the cards.

I'm a little apprehensive about posting this to many news groups for  
fear of the "stop cross posting rants"...

Thanks,
Jeff.


On Sep 2, 2009, at 8:10 AM, Howard Butler wrote:

>
> On Sep 2, 2009, at 9:28 AM, Mateusz Loskot wrote:
>
>> Jeff Hamann wrote:
>>> Mateusz,
>>>
>>> I noticed the migration of the liblas project from Subversion to
>>> Mercurial. Can we still check out the code using svn?
>>
>> Jeff,
>>
>> I'm not sure. Hobu?
>>
>
> no.  I wanted to make it easy for folks who've pulled libLAS behind  
> the wall to be able to still toss patches back over it (govt types,  
> especially).  hg does this for us in a way that svn doesn't.  I will  
> admit I still don't know my way around hg very well though :)
>
>>> Also, I wondered what's the status of integrating the liblas code
>>> into the gdal/ogr project?
>>
>> There is no such status as it has never been planned.
>
> Not currently planned, though frequently wished for.  The biggest  
> stumbling block is what does LAS data mean in the context of GDAL/ 
> OGR?  Is it vector (pure point) data with attributes, point patch  
> data ala Oracle Point Clouds, interpolated raster data ala DEM, TIN  
> data?  None of these concepts map into GDAL/OGR very well except for  
> the raster DEM, and getting to that point is something that is not  
> easily automated.
>
>

Jeff Hamann, PhD
PO Box 1421
Corvallis, Oregon 97339-1421
541-754-2457
jeff.hamann[at]forestinformatics[dot]com
http://www.forestinformatics.com