[vtkusers] vtkLandmarkTransform alghoritm

[David Gobbi]

The title of the paper advertises (states) that the method is closed form
(expressed as a set of equations that can be directly evaluated to give the
solution).

Within the paper, one important step of the method is done with a standard
iterative (i.e. not closed-form) algorithm to find the eigenvalues of a 4x4
matrix.

So, the solution presented in the paper is not closed-form.  The paper
states that a closed-form solution for the eigenvalues of a 4x4 matrix
exists, but it does not use the closed-form solution. It uses an iterative
solution instead.

It wasn't until later (maybe 2002?) that someone finally published a fully
closed-form solution.  But VTK uses Horn's method, not the new method.
 People who are just using the method (rather than implementing it)
probably will not care one way or the other... the two methods would give
nearly identical answers.

 - David


On Tue, Dec 6, 2011 at 3:17 PM, andon <antoniod987 at gmail.com> wrote:

> I've not understood what do you mean for " it isn't a closed form solution
> as
> advertised." and "The complete closed-form solution was developed later.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.vtk.org/pipermail/vtkusers/attachments/20111206/0328de5a/attachment.htm>