[Insight-users] destinguishing sheets from fibers

[lynx.abraxas at freenet.de]

Hello!


Is it possible to destinguish sheets from fibers by a filter (chain)?
The  fibers are ideally abritarily bend cylinders of diameter d and the sheets
are ideally abritrarily bend planes with a thickness d as well. The fibers and
sheets are randomly distributed in 3D space.

I  was  thinking  if this can be done by a 3D FFT, finding the local maxima in
the FFT-amplitude image, masking these with a local or  global  threshold  and
then doing a inverse FFT of only what is within the mask.
Would  this work? Would the sheets yield a stronger signal in the FFT than the
fibers?
Is there any other way to do this?
Is there perhaps some kind  of  skeletonization  filter  that  would  make  it
possible  to destinguish parts from the skeleton that belong to the sheets and
parts that belong to fibers?

What approach would be best if d is not exact but varies?


Thanks for any help or hints.
Lynx