[Insight-users] multiple inputs and different size

[Rashindra Manniesing]

HI,

Any advice on the following would be highly appreciated:

I am writing a class derived from ImageToImageFilter which has two
inputs and one output. One input and the output have the same properties
(dimension, sizes, pixeltype), the other (optional) input has a
different pixeltype and a different size (in x direction). It looks
something like this (the optional second input can be set by public
member functions)

template <class TInputImage, class TOutputImage>
class ITK_EXPORT A:
        public ImageToImageFilter<TInputImage,TOutputImage>
{
	// input same size/dim as output
	// Input/OutputPixelType is real

	// Second input, different size/pixeltype
        typedef Image<std::complex<InputPixelType>, ImageDimension>
SecondImageType;
        void SetSecondImage(const SecondImageType *);
        const SecondImageType * GetSecondImage(void);

}

Input2 is the fourier transform of input1 and internally, after the
required calculations, the result (image) will be backtransformed to
real domain (using itkFFTWComplextoReal ..). 

My questions:

- Because itkFFTWComplexToReal already takes care of the different
input/output size and pixeltype, is it still required to implement
the functions GenerateOutputInformation() and
GenerateInputRequestedRegion() ?

- Am I correct by stating that if at least one of the (input/output)
images has a different size than another one, these functions should be
implemented?


Many thanks for your help,
Rashindra Manniesing.