Non rigid .. RE: [Insight-users] about register point set with surface

[Luis Ibanez]

Hi Robert,

Deformable registration among PointSets may be prone to association
mistmatch. In particular if you use a association critrion such as
"closest point".

The challenge with using a DeformableRegistration in order to evaluate
the results of a Rigid registration is that Deformable registration is
much more complex than a Rigid registration. Therefore, you will never
know if the values that you obtain in the evaluation are the result
of your DeformableRegistration being poor, or they are the result of
your RigidRegistration being poor.

I would suggest you to simply use the residual distances directly
from the distance map. Just as you propose in your email.



    Regards,



        Luis



------------------------
Atwood, Robert C wrote:

> 
> This question got me thinkng about my registration project that I 
> mentioned in my previous question; quantifying the mismatch in 
> successive manufacturing stages using point-scan data. Before launching 
> into a whole new set of the registration classes and getting into the 
> 'supervisor wondering' mode ;-) I though I would as the advice of those 
> people more familiar with the mathematics of the methods.
> 
> If I take the two point sets, remove the rotation adn translation 
> mismatch (either by rigid registration or some other method), apply a 
> *Non-rigid* registration method such as suggested below, and then 
> extract the vector field, will this be a valid representation of the 
> degree of misfit between the two objects? My original idea was to 
> measure the value of the distance map of one point set at the points of 
> the other set. Would this just give the same result? Has anyone tried 
> something like that already?
> 
> 
> 
> Thanks for any advice
> Robert
> 
>  
> 
> -----Original Message-----
> From: insight-users-bounces+r.atwood=imperial.ac.uk at itk.org on behalf of 
> Luis Ibanez
> Sent: Fri 5/13/2005 1:52 PM
> To: Yixun Liu
> Cc: insight-users at itk.org
> Subject: Re: [Insight-users] about register point set with surface
> 
> 
> Hi Yixun,
> 
> I would suggest you to take the Surface and use all its points
> in order to create a Distance Map using the DanielsonDistance filter.
> 
> Then perform registration using the PointSet and the Image distance map.
> 
> You will find classes in ITK for performing PointSet to
> Image registration:
> 
> Follow the trail from:
> http://www.itk.org/Insight/Doxygen/html/classitk_1_1PointSetToImageRegistrationMethod.html
> 
> 
> Regards,
> 
> 
>     Luis
> 
> 
> 
> ---------------------
> Yixun Liu wrote:
> 
>  > Hi,
>  > 
>  > I have a surface represented by point set and another surface
>  > represented by mesh. I need to regeister them non-rigidly. Anybody give
>  > me some advice about it?
>  > 
>  > Thank you in advance!
>  > 
>  > Best regards,
>  > 
>  > Yixun Liu
>  >
>  >
>  > ------------------------------------------------------------------------
>  >
>  > _______________________________________________
>  > Insight-users mailing list
>  > Insight-users at itk.org
>  > http://www.itk.org/mailman/listinfo/insight-users
> 
> 
> 
> _______________________________________________
> Insight-users mailing list
> Insight-users at itk.org
> http://www.itk.org/mailman/listinfo/insight-users
> 
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> Insight-users mailing list
> Insight-users at itk.org
> http://www.itk.org/mailman/listinfo/insight-users