[Insight-users] Mutual Information rotation & translation and the metric value

[Seniha Esen Yuksel]

Hi,
I am trying to write a mutual information registration code (Viola Wells) 
that implements both centered rotation and translation. Therefore, I tried to combine
the examples ImageRegistration2.cxx (mutual information using translation transform) and ImageRegistration5.cxx  (centeredrigid2D transform using MeanSquaresImageToImageMetric). However, when I run my code to register the given images
BrainProtonDensitySliceBorder20.png and BrainProtonDensitySliceR10X13Y17.png, the angle 
values are extremely high (-3007.19) and I couldn't find where the problem is. 
 
Another question is, if I understood correctly, in these examples, we are taking
the final registration parameters as the best values. However, the last metric value
is not necessarily the best metric value, ie. we are not looking for the best metric 
that maximizes the mutual information but only taking the last one when we do "finalTransform->SetParameters( finalParameters )".  Please correct me if I misunderstood this part or I would appreciate if you could explain why it was not done as such or how I can do it.
 
 
Thank you very much,
Best regards,
Esen



Seniha Esen Yuksel 
Research Assistant 
Computer Vision and Image Processing Lab, 
Lutz Hall Rm 414, 
University of Louisville, 
Louisville, KY 
work ph: 502- 8521528 
home ph: 502- 8549856 
email: esen at cvip.uofl.edu // esenyuksel at ieee.org 
web: http://www.cvip.louisville.edu/~esen/

__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://public.kitware.com/pipermail/insight-users/attachments/20050713/da9c4cfe/attachment.html