[ITK] finite differences approximation & MultipleValuedCostFunction

[Francois.GIRINON at ensam.eu]

Le 09/08/2016 à 15:34, Dženan Zukić a écrit :
Hi François,

have you seen this<https://itk.org/Doxygen/html/group__FiniteDifferenceFunctions.html> list?

Regards,
Dženan

On Tue, Aug 9, 2016 at 7:20 AM, <Francois.GIRINON at ensam.eu<mailto:Francois.GIRINON at ensam.eu>> wrote:
Hello,

I wondered if ITK already included some methods to compute finite differences approximation and in particular for MultipleValuedCostFunction ?

Thanks for any help,
François.
_______________________________________________
Community mailing list
Community at itk.org<mailto:Community at itk.org>
http://public.kitware.com/mailman/listinfo/community


I've actually seen this, but what I'd like to do is computing for the  1 ... i parameters of a transformation :
df/dpi = (f(pi + epsilon) - f(pi + epsilon))/ (2*epsilon)

where :

  *   f is a MultipleValuedCostFunction
  *   pi , the ith parameter
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/community/attachments/20160809/7d0732a4/attachment.html>