Comment: Interesting point cloud registration publication

[tischi]

Hi,
just in the (unlikely) case you don't know it yet.
I attached a paper that the SPIM people at EMBL like in terms of point cloud registration.
Cheers, Christian

CPD_PAMI.pdf

[StephanPreibisch]

Hi, thanks @tischi .... I do know about it, it is actually great and I always wanted to add it as an implementation. Thanks for reminding me, I might just do it now :)

[StephanPreibisch]

I think it is way better than the relatively trivial ICP ...

[tischi]

I would really appreciate if you added it!
Probably (I hope) this would help for aligning time-series with nuclei that only correspond in terms of global pattern and not in terms of local similarities (i.e., in cases where the imaging is slower than cell migration and/or cell divisions). What do you think?

[StephanPreibisch]

If the drift is less than the average distances between nuclei, yes. But I do not expect it to be much better than ICP, but only one way to find out I think.