[Insight-users] 答复:About the metric value

Luis Ibanez luis.ibanez at kitware.com
Fri Dec 3 12:51:31 EST 2004


Hi Carrie,

You *must* read the ITK Software Guide.

      http://www.itk.org/ItkSoftwareGuide.pdf

in particular Chapter 8, "Registration" in pdf-pages 241-340.

The ImageMetrics in the ITK registration framework measure how
similar the Fixed image is to the transformed Moving image.

There are many different metrics available in ITK  (and you can
always add your own). For a full list of the current metrics,
please look at the "Groups"  documentation page

    http://www.itk.org/Insight/Doxygen/html/modules.html

and to the group "Similarity Metrics":

http://www.itk.org/Insight/Doxygen/html/group__RegistrationMetrics.html


Some metrics are optimal when they reach a minumum value, while some
other metrics are optimal when they reach a maximum value.  For example,
Mutual Information is optimial when it is maximized, while MeanSquares
Sum is optimal when it is minimized.  If you look at the Doxygen
documentation page of each metric you will find the condition under
which each metric is considered to be optimal.

This is the reason why ITK Optimizers allow you to switch between
minimization and maximization by using the methods MaximizeOn/Off
MinimizeOn/Off.


Mutual Information is known for producing very noisy output. You
will see this on the Metric plots of the ITK Software Guide.
You can reduce noise by smoothing the images previous to registration.
You can also use the Mattes implementation of Mutual information,
which is less noisy than the Viola-Wells implementation.  you will
find examples of both in the ITK Software Guide, and you will find
the associated source code examples in the directory


            Insight/
                 Examples/
                        Registration




Regards,


   Luis



------------------------------------
Carrie wrote:

> Dear All,
> 	I want to use metric value to stop my registration process. But
> I do not know clearly meanings of metric value in ITK. In my test, I
> found the metric value changed widely with different images. So I want
> to ask:
> 1. how to select a given metric value to stop the regitration process?
> 2. Does the bigger metric value mean the better result?
> 3. What's the meanings of metric value?
> 
> PS: part codes as follewed:
>   registration = itk.itkImageRegistrationMethodF3F3_New()
>   imageMetric  =itk.itkMutualInformationImageToImageMetricF3F3_New()
>     transform  =	itk.itkQuaternionRigidTransform_New()
>     optimizer
> =itk.itkQuaternionRigidTransformGradientDescentOptimizer_New()
>     interpolator = itk.itkLinearInterpolateImageFunctionF3D_New()
> 
> Regards,
> All
> 
> _______________________________________________
> Insight-users mailing list
> Insight-users at itk.org
> http://www.itk.org/mailman/listinfo/insight-users
> 
> 





More information about the Insight-users mailing list