[Insight-users] Translation-then-Rotation transform?

Luis Ibanez luis.ibanez at kitware.com
Fri Feb 18 01:56:47 EST 2005


Hi Zach,

We have planned to rework the transforms in order to make them
provide center of rotation but without making it part of the
array of values to be passed to the optimizer.

This modification require some care, in order to make sure that
we provide backwards compatibility.

At this point, the simplest thing to do is for you to create
a variation of the CenteredRigid2DTransform, where you don't
make the center part of the parameters array.

Please let us know if you need any help making those changes,
we will be happy to provide you some support.


    Regards,


       Luis


-------------------------
Zachary Pincus wrote:

> Hello, (and sorry to be so constantly emailing the list lately!)
> 
> I'd like to use a spatial transform that translates a point to a new 
> space and then performs a rotation in that new space about the origin. 
> Basically, I've got a moving image that I want to translate around a 
> fixed image with the optimization framework, and rotate about its 
> center. I'm not using the registration framework here; in reality I'm 
> querying trying to find an optimal parameter fit of a PCA shape model to 
> an image. So the "moving image" is really a PCA shape model.
> 
> Using the CenteredRigid2DTransform is a bit painful because it allows 
> the optimizer to change the translation and center of rotation 
> independently, whereas I have prior knowledge that the center of 
> rotation should *always* be about the origin in the new space.
> 
> Now, I could make a different version of the Rigid2DTransforma and 
> Euler2DTransform classes that would have this behavior (currently, they 
> rotate-then-translate). Is there an easier method, though? Or would I 
> really be better off for some reason using the CenteredRigid2DTransform?
> 
> (I've read the chapter about centered transforms in the software guide, 
> but I don't see why optimizing over a 5-dimensional parameter space is 
> any better for me in this case than over the more natural 3-dimensional 
> <translate_x, translate_y, rotate> space.)
> 
> Thanks,
> 
> Zach
> 
> _______________________________________________
> Insight-users mailing list
> Insight-users at itk.org
> http://www.itk.org/mailman/listinfo/insight-users
> 
> 





More information about the Insight-users mailing list