Useful range of values for step, spike, and standard deviation.
Vance Green
Hello,
I am geosciences grad student at the University of Arkansas. My thesis involves using python to automate the selection of ideal parameter combinations in LASground for accurate classification of airborne LIDAR data. The script creates multiple bare-earth classifications of a single LAS tile using a range of parameter combinations in LASground and then compares them using LASdiff to a manually-classified control surface. The chosen tile is meant to be representative of the entire study area, therefore the rest of the area can be classified using the “ideal” parameters.
Currently, my script takes a range of values for step, spike, and stddev and creates a new classification for each possible combination. It then compares each classification to the manually-classified control surface using LASdiff. Finally, it outputs a CSV file with an entry for each new classification containing file name, step, spike, and stddev values, and the number of points that differ from the control surface.
Up until this point, I have just used small ranges of integers for each parameter for testing purposes. I am now ready to attempt use a range of floats and conduct a full test, but I am struggling to determine an appropriate range for each parameter. I have scoured this google group and I have read the lasground_readme file. I have also read Agata Klimczyk’s research at Ludwig Boltzmann Institute (http://lbi-archpro.org/als-filtering/lbi-project/results/lastools). Agata has published useful guidelines that are especially helpful for determining a useful step size range, but I am still struggling with a range for spike and standard deviation.
Does anyone have suggestions for a useful range of values for step, spike, and standard deviation? I realize this is highly dependent on the terrain in question, but for this project a more general range is wanted to allow the script to narrow down the results.
Thanks,
Vance Green
