[Liblas-devel] chopping up .las

[Howard Butler]

On Dec 16, 2010, at 2:11 PM, Peter Tittmann wrote:

> Greetings all,
> 
> I wanted to see if someone could advise as to the most efficient use of LibLAS for the following:
> 
> I have a series of 200-500mb las files with classified points.
> 
> I have  a program (C++) that reads in .txt files  and doesn't like them bigger than about 10mb. I have written some python to batch the C++ app for multiple tiles. Obviously, using LibLAS to load the .las files directly into the app would be best but due to time/resource constraints thats not going to happen. 

I have just added the ability for lasblock to output .las files using the --write-points option.  Because of the way the chipper works, we can't filter *and* chip the data at the same time, so you'll have to do either the filtering or the chipping first, and then the other.  

> 
> I need to subset the large tiles spatially and by return #/classification. My idea at the moment is to use write a script to batch las2las2 to produce text files for each combination of spatial extent and point class, then use the python api to produce text files that are digestible by the app. 

las2las2 is gone.  It's all just las2las now.  The previous incarnation is called las2las-old, which retains the old calling conventions and arguments if you need it.

> 
> My question (finally), is weather this is the best way to approach this problem from an efficiency standpoint. I have used the python api to read through the points in the original (large) .las files and spit out text files with my criteria but its very brute force and slow. 

Using the above new functionality:

lasblock my_500_mb_file.las --capacity 1000000 --write-points million_point_file.las

#!/bin/bash
for i in $(ls million_point_file*.las)
	las2las $i $i-filtered.las --my-filter-options 
done

Hope this helps,

Howard