Segmentation/Subsample based on given given octree

[c.hess]

Hi,
i'd like to receive a segmentation/subsample of an object out of a very "noisy" point cloud. Noisy, because I scanned trees (no leaves, fine branches) in a forest surrounded by other trees. Beside the scan of the "noisy" tree-group I cut the trees and did a clean scan from each single tree outside the group.

My idea: Due to point picking I align/registrate the trees from both scans by picking prominent points in each point cloud (overlay of trees).
Than I compute an octree on the single, clear tree point cloud.
My Question: Can I switch this octree-structure to the other "noisy" point cloud to use it as a mask or 3D-shape-structure for a subsample?

[daniel]

Good idea... but you can't ;)

You could sample one point per cell of the octree, then compute the distance from the second cloud relatively to this sampled cloud and keep only the points close to it (it would be equivalent to a kind of "spherical" version of the octree ;). Of course you could also directly compute the distance between both clouds.