[Insight-users] Is this a wrapping bug ??

[Charl P. Botha]

On Fri, 2004-03-26 at 20:18, Rodrigo Trujillo wrote:
> Hi Charl,
> 
> 
> more methods whitch don't exist in wrapping to US2 and US3.
> 
> - 'GetDifferenceFunction'
> - 'GetElapsedIterations'
> - 'SetDifferenceFunction'
> - 'SetManualReinitialization'
> - 'SetMaximumRMSError'
>  - 'SetNumberOfIterations'
> - 'SetRMSChange'
> - 'SetState'
> - 'SetStateToInitialized'
> - 'SetStateToUninitialized'
> 
> apparently, the problem  is in class 'itkFiniteDifferenceImageFilter', which
> has only this class in Python:
> 
> - itkFiniteDifferenceImageFilterF2F2
> - itkFiniteDifferenceImageFilterF3F3
> - itkFiniteDifferenceImageFilterUS2F2
> - itkFiniteDifferenceImageFilterUS3F3
> 
> Is this the problem and what did you want ?

Yes exactly.  This can be fixed by adding the missing types to the
itkFiniteDifferenceImageFilter wrapping specification.

However: I don't know why the CurvatureAnisotropicDiffusionImageFilter
is wrapped for US2 and US3 in the first place.  From the documentation
at
http://www.itk.org/Doxygen/html/classitk_1_1CurvatureAnisotropicDiffusionImageFilter.html

Inputs and Outputs
        The input and output to this filter must be a scalar itk::Image
        with numerical pixel types (float or double). A user defined
        type which correctly defines arithmetic operations with floating
        point accuracy should also give correct results.

I would actually suggest that the US2US2 and US3US3 wrappings for the
curvature anisotropic filter are removed, as they help bloat the code
(the libraries are already huge) and shouldn't really be used.

What do you think?

Regards,
Charl

-- 
charl p. botha http://cpbotha.net/ http://visualisation.tudelft.nl/