[Insight-users] VectorImage Registration Metric

[Luis Ibanez]

Hi Kevin,

I put the current draft of the Vector registration classes in the
NAMIC Sandbox:

          http://www.na-mic.org/svn/NAMICSandBox/trunk/

under the directory:   VectorImageRegistrationMethod

http://www.na-mic.org/svn/NAMICSandBox/trunk/VectorImageRegistrationMethod/

Several changes are required in ITK for getting this to compile.

The basic changes are in

    VectorInterpolateImageFunction

that was assuming a return value of FixedArray< RealType, Dimension >
instead of building the type from the RealType of the image pixel type,
which in this case is expected to be a vector type.

A second change is that we didn't have any NumericTraits defined for
FixedArray.


These two changes have secondary effects that I'm still ironing out
before they can be committed in to the CVS repository.


---

The metrics in the NAMIC sanbox also need some massaging.

The Gradient of the vector image must be computed as a matrix
and multiplied with the Transform jacobian in order to produce
another matrix, that then will be collapsed into a gradient
vector. (to be done)....




     Luis



-----------------------
Kevin H. Hobbs wrote:
> On Fri, 2008-05-02 at 15:30 -0400, Luis Ibanez wrote:
> 
>>Hi Kevin,
>>
>>Thanks for trying this.
>>
> 
> 
> No problem.
> 
> 
>>I will give it a try to draft a customized image metric during the
>>weekend and will get back to you.
>>
> 
> 
> Thank you.
> 
> 
>>Would MeanSquare be the best metric ?
>>
> 
> 
> It would be totally sufficient, I can't say that any metric would be
> "best".
> 
> 
>>Would you need a scaling factor between components ?
> 
> 
> I don't believe _I_ would. These cells were injected with enough dye
> that they saturate both sensors in many places.
> 
> I don't know enough about the dyes, lasers, filters, and sensors to say
> that a difference in intensity should or should not be scaled.
>