[Insight-users] Using Rigid3dPerspectiveTransform as a filter?

[Emma Ryan]

Hi,

  Has anyone attempted 2d/3d registration using itk ?

 How does one use Rigid3DPerspective transform ? 

Typically, a transform, such as translation or affine is defined as the Transformtype to be used by the registration algorithm. 

 If the input to perspective transformation is 3D data and output is 2D, then should the registration occur in 2d ?
Obviously the registration is to occur in 3D. But how does the optimizer do a similarity check on a 3D and 2D input ?

What is the data flow like for 2d/3d registration using itk ?

Is it :

Model 1 :

Load 3d volume -> Apply perspective transform - > Use the generated DRR to 'similarity match'  with the reference 2D image ?

So then, how does one apply the perspective transform ? Is the registration process now a  2d optimization problem?.  This would be incorrect.
or 

Model 2

Load 3d volume -> Start Registration - > PerpectiveTransform automatically generates DRR and compares with the reference 2D image 

How does this process happen ?


any clues ?

thanks in advance,
Emma



   





       
____________________________________________________________________________________
Choose the right car based on your needs.  Check out Yahoo! Autos new Car Finder tool.
http://autos.yahoo.com/carfinder/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://public.kitware.com/pipermail/insight-users/attachments/20070904/506b4e59/attachment-0001.htm