[Insight-users] how can I shorten the registration time?

[Luis Ibanez]

Hi Carrie,

You can register this size of images in less
that 2 minutes in a Pentium 4 at 2Ghz.

What you want to do is to subsample the images
at a quarter of resolution, register them that
way. Then use the resulting transform for
initializing a registration at half resolution.

Note that a simple use of the ResampleImageFilter
with nearest neighborhood interpolator may be
the best combination for subsampline when you are
looking for ways of improving performance.

Simply skip the Gaussian smoothing since it adds
too much time to the preparation for the registration.


    Regards,


      Luis


--------------
Carrie wrote:

> Hi all,
> 
>          I write some sentences for image registration with python. The 
> extends of fixed and moving images are about 180*216*180. Some codes as 
> followed:
> 
> /        registration = itk.itkImageRegistrationMethodF3F3_New()/
> 
> /        imageMetric  = 
> itk.itkMutualInformationImageToImageMetricF3F3_New()/
> 
> /             transform     =         itk.itkQuaternionRigidTransform_New()/
> 
> /        optimizer    = 
> itk.itkQuaternionRigidTransformGradientDescentOptimizer_New()/
> 
> /        interpolator = itk.itkLinearInterpolateImageFunctionF3D_New()/
> 
>  
> 
> If I wanted the satisfied registration result, it need run a long time 
> above half an hour.
> 
> Now I want to ask whether it is normal. And are there any methods to 
> shorten the running time?
> 
>  
> 
> Regards
> 
>                                                                         
>                   carrie
> 
>         
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> Insight-users mailing list
> Insight-users at itk.org
> http://www.itk.org/mailman/listinfo/insight-users