Subsampling and Alignment Algorithms

[phavel]

Bonjour Monsieur Girardeau-Montaut,

awesome software! It's perfect for working with big pointclouds (in my case ~90 Mio points) for georeferencing and surface reconstruction! :) (Meshlab always crashes in these cases and couldn't handle UTM coordinates)

I'd like to ask you two questions and would be very grateful if you could help me:

Which algortihm do you use for the simple spaial subsampling?
And which algorithm do you use for the alignment of two pointclouds by picking point pairs?

I'm using CloudCompare for my master thesis and I would be very grateful if you could help me with references or papers if you have used special algorithms or even the names of the algorithms if they are known.

Many thanks et le bonjour de près de la frontière, de Trèves
Patrick

[daniel]

Hi

Well the subsampling algorithm is very simple (I don't even know if it has a name or an associated reference): simply pick a random point, then flag all the points around it as to be removed (i.e. inside a sphere - the radius is specified by the user). Repeat the process until all points have been either picked or flagged.

The point-pair-based alignment is done with the Horn rigid 3d registration method: http://people.csail.mit.edu/bkph/papers ... tation.pdf

Tschüss,

[phavel]

Many thanks for the quick answer! :)
Okay perfect. I've understood the subsampling algorithm now and thanks a lot for the paper for the alignment algorithm.
You helped me a lot! :)

Kind Regards
Patrick

[imdly]

Hello, do you find the related papers for the spatial subsampling? And the detailed introduction? I also write a paper, but I don't know how to introduce the method and the reference paper.

[daniel]

I don't think there's any official paper explaining this algorithm.

But you have the explanation above (and probably one or two older messages on this forum). And also the source code that might be referenced? (https://github.com/CloudCompare/CloudCo ... s.cpp#L695)

[imdly]

It helps. Thank you very much!