[Insight-users] Versor3DRegistration and the masks of meanSquareMetric

[Karthik Krishnan]

Martin Urschler wrote:

> Karthik,
>
> Thanks for your answer.
> Concerning the VersorRigid3DTransform:
>
> If I have two images with rigidly transformed anatomical structures 
> and I don't know the rotation center of the transformation (I only 
> know the center of mass from the CenteredTransformInitializer, which 
> is definitely not the desired rotation center) then the 
> versorRigid3DOptimizer will lead me to a local minimum of transform 
> parameters that is still around the same rotation center? If this is 
> the case then this transform is only useful for me for small rotations 
> which might be approximated as a translation, right?
>
> In this case, wouldn't it be useful to write a 
> CenteredVersorRigid3DTransform, that optimizes the rotation center as 
> well? Is this already planned? I could try it if you could point me to 
> some more thorough explanations of the Centered vs. NotCentered 
> transforms...

I may be wrong here, but is there any reason to optimize both the center 
and the translation ...

For the VersorRigid3DTransform which may be re-written as

p' = R(p-C) + C + T = Rp +  C(I-R) + T

If you were to optimize just R and T and specify a default center, you 
should still be able to arrive at an equivalent translation that rotates 
about any other center, if the optimizer does a good job.


>
> regards,
> Martin
>
>