[Insight-users] Registration with a deformation field

[Luis Ibanez]

Hi Andreas,

     2 Minutes per iteration of the optimizer
     is way too long for a typical registration.


In a 2D image of size 256x256,  this should
take just fractions of a second per iteration.


You may be using this code compiled for "Debug" mode,
which seriously affects its run-time speed.

Please make sure that you recompile the code for "Release"


     Regards,


             Luis



-------------------------
On Fri, May 14, 2010 at 10:26 AM, Merrem, Andreas
<andreas.merrem at medma.uni-heidelberg.de> wrote:
> Hello,
>
> I tried to implement a registration algorithm in the itk framework with a
> deformation field as a transform for a 256X256 2D image.
>
> This field should be displacement vector for each pixel using all image
> pixels as grid points (i.e. 2*number of pixels degrees of freedom) and a
> gradient descent optimizer. It worked with the B-spline transform class, but
> it was really slow (2 minutes per optimizer iteration). I'm not sure if that
> was because B-spline class somehow makes the program compute splines for
> interpolation although there are no points in between the grid points. Does
> anyone know a faster way of implementing a deformation field?
>
>  Thank you for your help,
>
> Andreas
>
>
> _____________________________________
> Powered by www.kitware.com
>
> Visit other Kitware open-source projects at
> http://www.kitware.com/opensource/opensource.html
>
> Kitware offers ITK Training Courses, for more information visit:
> http://www.kitware.com/products/protraining.html
>
> Please keep messages on-topic and check the ITK FAQ at:
> http://www.itk.org/Wiki/ITK_FAQ
>
> Follow this link to subscribe/unsubscribe:
> http://www.itk.org/mailman/listinfo/insight-users
>
>