[elastix] Large translation errors

Rupert Brooks rupert.brooks at gmail.com
Mon Jun 27 13:53:49 CEST 2011


Hi Manisha,

I have done some similiar experiments in the past.  I didnt use Elastix, but
a similar approach should work - generate the mask of pixels of interest,
and then feed that to the registration process as a mask.  However, I found
that the results were extremely dependent on the scale of the features used.
 For what its worth, these experiments and results are described in Chap 6
of my thesis.

http://www.rupertbrooks.ca/downloads/Brooks_PhDThesis.pdf

However, since that time Benhimane et al took this idea in a somewhat
different direction.  Ultimately, i think the approach to this problem
described in their paper

 S. Benhimane, A. Ladikos, V. Lepetit, and N. Navab. Linear and quadratic
sub-
sets for template-based tracking. In Proceedings of the Conference on
Computer
Vision and Pattern Recognition (CVPR'07), 2007

is quite insightful and probably a better one.  As far as i can see, you
could generate a mask with their approach, and feed that mask into Elastix -
however, i have not actually tried this, it just seems to me it should work.

Good luck,

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



On Fri, Jun 24, 2011 at 02:19, manisha rajdeep
<manisharajdeep at yahoo.co.in>wrote:

> Hi,
> I am trying to register satellite images having on average 200 pixel shift.
> Size of the images are around 7K*7K
> I am able to register many of the images with as much 10 resolution levels
> (pyramids).
>
> I am trying another solution like instead of Random sampling,can i give
> corner or edge features of images like output of Harris corner detection as
> input. Is it possible through masking,any help in this area will be
> beneficiary for me.
> Plz suggest the appropiate way.
>
> _______________________________________________
> elastix mailing list
> elastix at bigr.nl
> http://lists.bigr.nl/mailman/listinfo/elastix
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.bigr.nl/pipermail/elastix/attachments/20110627/325af2f9/attachment.html>


More information about the Elastix mailing list