CloudCompare forum

Hi Daniel,

One of my goal is to align two point cloud using ICP algorithm in batch (it is why I am using the command line mode).
To do that, I think I need that my clouds are close to each other, therefore I apply a match center (MATCH_CENTERS).
Then, I load the new cloud and the reference cloud to apply ICP... So far so good.
But finally, I need to apply exactly the same transformation to a third point cloud (using APPLY_TRANS). There is no problem for my second transformation made with ICP as this function output a transformation matrix, but I have not the transformation (translation) matrix following the match center.
Eventually, my questions are:
Do you think that my first step (MATCH_CENTERS) before applying a ICP is necessary ? (I think so...)
If yes, is there a way to output the transformation matrix following the match center ?
Or better, do you have any solution to make both of my transformations (MATCH_CENTER and ICP) without output the intermediate cloud and export the transformation history (as it is possible with the interface mode) ?

Thanks a lot,

Olivier

Hi,
I rekindle the post because I didn't find any solution. I realize that maybe my post was not really clear (excuse me for my English).
I just want to know if it is possible to extract the transformation matrix of a MATCH_CENTER in command line mode ?
Or, is it possible to apply several transformations on a point cloud (e.g MATC_CENTER + ICP) and output the resulting transformation matrix ?

Thank you so much

Sorry I just missed your initial post ;)

With the current version it's not possible. But we could indeed output the transformation just as the ICP one. Or more generically export the 'transformation history' as you suggest. I'll simply add your post to the TODO list and we'll see what comes out ;)

P.S.: the 'bounding-box center matching' is done automatically when performing ICP if and only if the clouds are very far apart.

Thank you for your answer.
What do you mean by very far apart, what is the threshold ? My scenes are at a distance from 20 to 50m is that enough for the automatically bounding box center matching ?
Thanks

I don't remember exactly but I believe it's when the bounding-boxes are disjoint (or maybe the octree's bounding-boxes).

ok, Thank you