[Insight-developers] ITKV4 kernel spline comment

[Rupert Brooks]

Nice improvement!

I am interested in merging - but perhaps the right answer is to merge
with ITK main.  Then both elastix and my code can depend on the ITK
version.

Hans - looks like you are working on the changes for the
ThinPlateSpline transform family.  Whats the best way to proceed here?

Rupert

--------------------------------------------------------------
Rupert Brooks
rupert.brooks at gmail.com




On Wed, Jun 8, 2011 at 08:06,  <M.Staring at lumc.nl> wrote:
> I second that!
>
> Rupert, I made some modifications to your code to increase the TPS performance at several places. You can check it out at
>
>  https://svn.bigr.nl/elastix/trunkpublic/src/Components/Transforms/SplineKernelTransform/
>
> using elastixguest as username & password. From my svn log:
>
> -PERF: many performance enhancements for the thin plate spline.
>  - QR decomposition for matrix inversion is much (x10) faster then SVD
>  - Huge speedup for Jacobian computation, using the facts that LMatrixInverse is sparse,
>    has a specific shape, and is symmetric.
>  - Cached decomposition of LMatrix, beneficial when SetParameters() is called (every registration iteration).
>
> Are you interested in merging with your code?
>
> Thanks for your IJ contribution,
>
> Marius
>
> -----Original Message-----
> From: insight-developers-bounces at itk.org [mailto:insight-developers-bounces at itk.org] On Behalf Of Rupert Brooks
> Sent: woensdag 8 juni 2011 13:49
> To: Insight Developers
> Subject: [Insight-developers] ITKV4 kernel spline comment
>
> Hi everyone,
>
> In the spirit of helping with the code review and cleanup I've been going through ITKV4 and trying to port old code over and review the new interfaces, particularly in the registration components.
>
> 1. I noticed in this patch for review
> http://review.source.kitware.com/#change,1551
>
> that someone is working on expanding the KernelSplines to include many different kernels beyond the 6 or so that were already included.
> Thats excellent!   Might i suggest that while we have the opportunity
> to change backwards compatibility, we flip the use of target and source landmarks, so that the Jacobian is feasible to calculate?  This is needed for use of these transforms in the registration framework.
>
> This is described in an insight journal article here
> http://www.insight-journal.org/browse/publication/145 Sorry for the self-promotion.
>
> I signed up to Gerrit, but it wasnt obvious to me how to comment on this change there.  Do i have to add myself as a reviewer to the change?
>
> Cheers
> Rupert
> --------------------------------------------------------------
> Rupert Brooks
> rupert.brooks at gmail.com
> _______________________________________________
> 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://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-developers
>