[Insight-users] point cloud registration

[Luis Ibanez]

Hi Dennis,

I will be interested in seeing your counter examples that
indicate that non-homogeneous scaling combined with
rotation is not equivalent to shearing.


About the ScaleSkewVersor transform... if you dont want
skew to happen, then simply set the skew parameters to
zero...


Otherwise, plan B, is for you to take the source code
of the Similarity3DTransform or the ScaleSkewVersorTransform
and modify it to suit your needs.  If you do so, we will
be happy to include your new transform into the toolkit.
Writing a transform is not too difficult. You mainly have
to work on the "TransformPoint" method and the "GetJacobian"
method.

Please let us know if you need any assistance in writing
these new Transform class.


   Thanks


      Luis



--------------------
Dennis Wenzel wrote:
> Hello Luis,
> 
>> Performing non-homogeneous scaling and rotations
>> is equivalent to performing shearing. Therefore
>> once you consider having non-homogeneous scaling
>> you can equally well go for a full Affine transform.
> 
> 
> Non-uniform scaling and rotations are not equivalent to shearing -- I 
> can think of counterexamples to indicate so.
> 
> Granted, all 12 parameters of the 4x3 submatrix (or 3x4 depending on how 
> you look at it) of the 4x4 matrix representing the affine transform will 
> be modified by allowing for non-uniform scaling, rotation and 
> translation. But, in the case of the Affine transform, I can see that 
> those parameters are solved for independently, so there is no way to 
> ensure that skew is not part of the solution when you optimize.
> 
>>
>>
>> You are right, the Similarity3DTransform was added
>> to ITK after release 2.0.1.  If you decide to use
>> this transform you may have to download the CVS
>> version of ITK.  Note that we anticipate to release
>> ITK 2.2 by August 31st.
> 
> 
> 
> I do not want any skew, so it seems that I need a variation of the 
> Similartiy3DTransform which allows for non-uniform scaling.
> 
>>
>>
>> If you want to insist in using non-homogeneous scaling
>> then you probably want to look at the transform:
>>
>> ScaleSkewVersor3DTransform
>> http://www.itk.org/Insight/Doxygen/html/classitk_1_1ScaleSkewVersor3DTransform.html 
>>
> 
> 
> This is closer to what I need, but it still allows for skew.  If I used 
> this in a registration, would I need a special optimizer?  Even so, how 
> can I prevent any skew from occurring during a registration?  How 
> difficult would be to write a ScaleVersor3DTransform?
> 
> Dennis
> 
>