[Insight-developers] congealing

[brian avants]

hi jim

after thinking about it a bit more , i believe the multivariate metric
might be usable for a congealing like method:

http://review.source.kitware.com/#/c/4807/
http://vis-www.cs.umass.edu/papers/congeal_3D.pdf

or the same thing could be done with itkObjectToObjectMultiMetricv4

it would still require a registration method, though.  what would happen in
that method would be ( loosely speaking )

*  loop over all images
  *  for the j^th image, add all other i images ( i != j ) to the
Multi-Metric such that you can compute
      \sum_i   Metric(  image_i = fixed ,  image_j = moving , T_j )
      which will drive image_j  to the group space
  *  store the gradient of T_j
*  update all transformations, check convergence, start again

anyway, just a thought.   most of the work would be in writing the
CongealingRegistrationMethod and it would be there that you decide what the
optimization domain would be and similar details.

brian
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.itk.org/pipermail/insight-developers/attachments/20120721/e53eb493/attachment.htm>