[Insight-developers] feature matching recomendations?

Stephen R. Aylward Stephen.Aylward at Kitware.com
Wed Jun 7 10:48:48 EDT 2006


One idea is that itk already implements this as spatialObject-to-image 
registration.

"features" can be considered "models."   We setup methods for sparsely 
sampling models when evaluating a metric.   Works quite well.   No need 
for another registration framework.   Could even define, for example, an 
"NJectFeatureSpatialObject" and specify the order of the feature (e.g., 
0 for intensity match, 1 for derivative match, etc) as well as the location.

Stephen


There are examples of this in the software guide.

Stephen

Miller, James V (GE, Research) wrote:
> If the plan is use the feature matching for registration, then I suggest
> adding a new metric to ITK.
> 
> Parallel to this, we could consider allowing the image metrics to have to 
> have a set of "samples" on which they are evaluated.  These samples
> could be randomly generated or prescribed by a feature detector on the 
> fixed image.
> 
> I would like to have ITK's registration framework expanded to include
> more feature based registration techniques.  All that should be needed
> is a new set of metrics.
> 
> Jim
> 
> -----Original Message-----
> From: insight-developers-bounces+millerjv=crd.ge.com at itk.org
> [mailto:insight-developers-bounces+millerjv=crd.ge.com at itk.org]On Behalf
> Of Richard Beare
> Sent: Tuesday, June 06, 2006 9:15 PM
> To: insight-developers at itk.org
> Subject: [Insight-developers] feature matching recomendations?
> 
> 
> Hi,
> 
> Just wondering whether anyone had recommendations for implementation
> of correlation type  feature matching. I'm interested in the standard
> type of thing - normalised cross correlation between a neighbourhoods
> centred on each  feature. There will be lots of pairs of features for
> which the correlation must be calculated. There seem to be a couple of
> options - adapting the registration metrics, using
> NormalizeCorrelationImageFilter which doesn't seem to be quite right
> for the purpose, or writing something new. Are there any other
> alternatives?
> 
> Thanks
> _______________________________________________
> Insight-developers mailing list
> Insight-developers at itk.org
> http://www.itk.org/mailman/listinfo/insight-developers
> _______________________________________________
> Insight-developers mailing list
> Insight-developers at itk.org
> http://www.itk.org/mailman/listinfo/insight-developers
> 

-- 
=============================================================
Stephen R. Aylward, Ph.D.
Chief Medical Scientist
Kitware, Inc.
http://www.kitware.com


More information about the Insight-developers mailing list