[Insight-users] Polar transforms
Jakub Bican
jakub.bican at matfyz.cz
Wed May 18 09:37:43 EDT 2005
Hi all,
i am implementing a multidimensional polar transform. It transforms first two dimensions of cartesian space into <angle,radius> coordinates, leaving the other dimensions unchanged (i.e. - multidimensional cylindric transform). I have implemented both forward and backward transforms. To transform space to cylinder in general position (to transform other than first two dimensions), it is neccessary to transform the image first by for example Euler3DTransform, or PermuteAxesImageFilter. To avoid excess memory and time, i implemented PermuteAxesTransform, which does the same as the filter, but it is a transform.
Now i have two possibilities what to do:
1) imlement something like "bulk transform" as it is in BSplineDeformableTransform into the polar transforms, or
2) implement general "ComposeTransformsTransform", which keeps a queue of transforms and successively transforms points and vectors by all of them. So it is possible to do several transforms in just one ResampleImageFilter.
I find the second solution more laborious but principally much more useful. But before i will do it, i need to know surely, if there is not such class in ITK yet, because i did not find it and i think that it is really useful and it should be implemented. Or is there any other way how to pass several transforms into one ResampleImageFilter??
I also found AzimuthElevationToCartesianTransform class in ITK. This does a kind of spheric transform and i must say that it is implemented badly. It uses several parameters to "sample" the "polar" space once more. All these parameters are redundant to properly set Size, Spacing and Origin parameters of ResampleImageFilters output and only comlicates the situation. Keeping in mind that transforms are transforming "real" coordinates, such transforms as polar, cylindric or spheric do not need any parameters. Everything is just question of proper sampling of input and output space.
Finally, i found several requests for polar transforms in insight-users archives, so if it is still open and if it is possible, i would be pleased to submit my classes into ITK.
Regards
Jakub
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://public.kitware.com/pipermail/insight-users/attachments/20050518/d4d457a3/attachment.htm
More information about the Insight-users
mailing list