[Insight-users] aligning two contoursets

[Karthik Krishnan]

On Tue, Feb 3, 2009 at 10:26 AM, Yogish Mallya <mallyayogish at yahoo.co.in>wrote:

> Dear Karthik,
>
> I have attached an image in the mail to give you further clarifications on
> the problem. I want to warp contour B on to contour A
>
> I gone through some of the publications on robust point matching. Looks
> like point matching technique may be overkill for my problem. I was
> wondering can't we use ordering of the points for both the contour ?  Is
> there any functionality to fit b-spline functions to these two contours and
> then evaluate at different points for estimating deformation ?
>

You can certainly fit a BSpline curve / bezier curve etc to the two point
sets.

The classes to do so are (vtkBezierContourLineInterpolator /
itkBSplineScatteredDataPointSetToImageFilter)

However, by volunteering to take the approach of sampling the two curves
you'll be assuming that the sample points do form a good point
correspondance:

You can then use TPS or one of the other kernel splines in ITK to find the
transform that maps .

-------

Are you sure RPM is an overkill for your problem ? I doubt if the algorithm
will mess up your point ordering.

RPM (vtkpxTPSRPMRegistration) class minimizes
- Distance between the deformed (TPS) points and the target point set
- magnitude of affine + magnitude of non-rigid component of the TPS
(regularization).

The latter will be high if point ordering is messed up (the curve loops over
itself etc).


> Thanks in advance
> Yogish
>
> --- On *Tue, 3/2/09, Karthik Krishnan <karthik.krishnan at kitware.com>*wrote:
>
> From: Karthik Krishnan <karthik.krishnan at kitware.com>
> Subject: Re: [Insight-users] aligning two contoursets
> To: mallyayogish at yahoo.co.in
> Cc: insight-users at itk.org
> Date: Tuesday, 3 February, 2009, 9:10 AM
>
> Yogish:
>
> You should take a look at the Robust Point Matching algorithm from Chui and
> Anand Rangarajan:
>  H.  Chui and Anand Rangarajan, "A new algorithm for non-rigid point
> matching", CVPR, 00
>
> It solves the alignment problem with both rigid and non-rigid variants for
> arbitrary number of source and target points, with point correspondances being
> unknown. It is also fairly resilient to outliers. If you are looking for readily
> available source code, its available as a part of the BioImageSuite, in the form
> of a VTK algorithm. Look for source files of hte form vtk*RPM* and produces a
> vtkTransform for you.
>
>
> -- Karthik Krishnan
> R & D Engineer,
> Kitware Inc,
> Ph:  518 371 3971 x119
> Fax: 518 371 3971
>
>
>
> Yogish Mallya wrote:
> > Hi,
> >
> > I have to write an application in which I need to align two contours
> (open)having different number of control points.  In other words one of the
> contours has more number of points compared to other. Is there any class in ITK
> to warp a contour on to another?
> >
> >  Best Wishes
> > Yogish
> >
> >
> > ------------------------------------------------------------------------
> > Unlimited freedom, unlimited storage. Get it now
> <http://in.rd.yahoo.com/tagline_mail_2/*http://help.yahoo.com/l/in/yahoo/mail/yahoomail/tools/tools-08.html/>
>
> > ------------------------------------------------------------------------
> >
> > _______________________________________________
> > Insight-users mailing list
> > Insight-users at itk.org
> > http://www.itk.org/mailman/listinfo/insight-users
> >
>
>
>
> ------------------------------
> Add more friends to your messenger and enjoy! Invite them now.<http://in.rd.yahoo.com/tagline_messenger_6/*http://messenger.yahoo.com/invite/>




-- 
Karthik Krishnan
R&D Engineer,
Kitware Inc.
Ph: 518 371 3971 x119
Fax: 518 371 3971
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.itk.org/pipermail/insight-users/attachments/20090203/058ca736/attachment.htm>