If you just want to assess the interpolation, then I'd suggest
something along the lines of repeatedly rotating an image multiple
times (say by 36 degrees) and compare the multiply interpolated
version with the original. See the evaluations in eg
Philippe Thévenaz, Thierry Blu, and Michael Unser. "Interpolation Revisited".
IEEE TRANSACTIONS ON MEDICAL IMAGING, VOL. 19, NO. 7, JULY 2000 pp 739-758
Best regards,
-John
On 25 November 2011 14:42, Gonzalo Romero <[log in to unmask]> wrote:
> Dear all:
> I'm trying to validate the interpolation methods used at the construction of
> my FDG-PET template. My strategy is:
> -register the image A to the image reference B using a interpolation method
> (e.g. trilinear) and obtain the matrix transform MT.
> -calculate the anti-transform matrix MT-1
> -apply this matrix MT-1 to the image registered and obtain an image
> resultant C (in theory, similar to A)
> -calculate an image error E like: E=A-C
> -calculate other error using a B-spline 7 like a goldstandar (image S)
> E=S-C
> I realy want to know if this method is valid to validate the interpolation
> methods.
> Best regards
> Gonzalo
> --
>
> Bioing. Gonzalo M. Romero Mascarell
> tel. +54 261 4285884
> móvil. +54 261 5782419
> P Si no es necesario no imprima este correo.
> Todos somos responsables por el cuidado del medio ambiente.
>
|