Correspondence for posterior model in shape completion

[Maia R.]

Dear all,

I'm a bit confused on how to find the correspondence between the model and the target points: do I need to find correspondence from model.reference or from model.mean  and the "partial" target to be reconstructed ?

In this https://codeshare.io/ZJLqkX from Elon (https://groups.google.com/g/scalismo/c/QRnhj_tL364) , he is looking the correspondence between the model.mean and the target and starts the non-rigid icp with the model.reference.

But in Tutorial 9, the non-rigid ICP is started with the model.mean ?

Could you explain what would be the difference ?

Thank you,

Best regards, Maia

[Andreas Morel-Forster]

Hi Maia

I think the following should be done:

- You start with the mean, as without having seen any data, it is your best guess.

- When looking for correspondence, then you always search the correspondence from your current state, so the moving mesh. This is under the assumption that the closer you are, the better you can estimate the correspondence.

I only had a quick look at the code and I would maybe replace the starting point with the mean instead of using the reference on line 64.

You are right for the initial visualization the correspondences are searched by starting from the mean. But later on in the ICP on line 59 it seems that the moving mesh is used for fitting as I would do it too.

Best, Andreas

--
You received this message because you are subscribed to the Google Groups "scalismo" group.
To unsubscribe from this group and stop receiving emails from it, send an email to scalismo+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/scalismo/20e548a0-ccdb-45e1-98f7-326906a275fdn%40googlegroups.com.

--

*****************************************

Andreas Morel-Forster
Maiengasse 12
CH-4123 Allschwil

MOB.: +41 76 494 0343

MAIL: a.morel...@gmail.com
*****************************************