[Insight-users] cross-correlation of two images

[kc]

Hello,

I would like to use itkNormalizedCorrelationImageFilter to find the
cross-correlation of two (2D) images.

 I have looked at the test code in itkNormalizedCorrelationImageFilterTest which
was helpful.  However, rather than using the AnnulusOperator I would like the
template to be an entire (2D) image.

Is it currently possible (using itkNormalizedCorrelationImageFilter or
otherwise) to find the cross-correlation between two images (as done between
the annulus operator and the image in itkNormalizedCorrelationImageFilterTest)?
 I want to do more or less exactly what is done in the test code above,
including writing out the correlation image, so
NormalizedCorrelationImageToImageMetric is not really what I am looking for.

To use itkNormalizedCorrelationImageFilter do I have to write a
NeighborhoodOperator that is basically just a rectangle of pixel data?

My main question is: what should I give to the SetTemplate() method of the
itkNormalizedCorrelationImageFilter if I want to use an entire image for my
template?

Lastly, from looking at previous posts to the mailing list I see that others
would like to have the same ability using the FFT (specifically in post:
http://public.kitware.com/pipermail/insight-users/2005-October/015296.html). 
Has anyone implemented a FFT based cross-correlation function like that
mentioned in the post above?

Any suggestions or pointers to example code would be greatly appreciated!

Thanks,
Kevin