Final SimpleTree Project Release - Update
I reuploaded version 4.33.03. I hope it is stable. I was able to process 25 plots with less then 24 hours over last night.
The script I used is named:
- an xyb file of a merged multiple scan forestry scene with full Faro (French NFI scanners) software filters applied. That means no visible occurence of either so called Ghost Points or Sky points included.
- the scene should be a ~15-20 meter radius scene. Feel free to carefully extend yourself this radius on your own. If you use larger plots it might still be better to subdivide and merge the plots.
- The target trees of my QSM modeling method are leaf-off trees. Unhandled leave or needle hits will cause the twigs to be largely overestimated. You can find a supervised machine learning step in my plugin, described here:https://www.youtube.com/watch?v=ZYdmWVxtPDE&t=344s. It is semi manual and cannot be build in the pipeline. You would need to: process a plot until you have segmented trees and save them (fully automatically). Follow the given tutorial to denoise them. Then import all denoised clouds and plugin my QSM step afterwards.
- In case you cannot afford to denoise I recommend you to stick only to the cylinders tagged with NO_ALLOM. Those are included in the ply files named “good”. The rest of the cylinders is less stable, e.g. statistically corrected and included in the ply folder “bad”.
- My topological order is known to choose incorrect paths. That leads to possible errors in the stem detection. Please assure to check the ply files in the folder color. Branch Order 0 (the stem) is colored in brown, the rest in green.
- Do not use georeferenced clouds, they should be centered roughly at 0,0,0. My plugin uses PCL and PCL is bounded to floats leading to problems with larger point coordinates.
- Stick to the citations given here: http://rdinnovation.onf.fr/projects/computree-simpletree-beta-version. By citing my SimpleTree paper, the Computree paper I recommend and PCL paper you at least give