I’ve got fairly high density TLS scans of a conifer forest, but I have a site with a LOT of small trees with low branches that overlap one another. Right now I’m JUST trying to get the stem position + dbh. I can clearly see the cylinders of the small trees, but the myriad branches are causing problems with the clustering. Any ideas? I’ve tried several versions of the Euclidean denoising and Remove noise from the stem seeds/slice (what is the difference) but I’m still left with a decent number of low branch clusters. Is it possible to filter out points that are not falling on cylindrical surfaces?
More to the point: what are the best practices for a lot of low branches + small trees?
Small note, the “Remove noise from the stem seeds” parameter does not seem to do anything – I get the same number of points removed regardless of the angle I put in (I’ve tried 15 to 75 degrees). Is there a bug in that step?