next up previous
Next: ``Homotopic_Thickening'' -- conditional Up: Topology Previous: ``Cavity_Deletion'' -- delete

 

``Homotopic_Thinning'' -- thin an input volume

 

purpose:
Thin an input binary-valued volume. ``Thin'' means to delete the outer layer of objects without changing the connectivity (homotopy) of the objects [24,25]. By thinning to completion, you can get the skeletons of objects in the volume.

The operation can be done in 3-D or 2-D. Also, when thinning to completion, short branches of the skeleton -- such as branches corresponding to small and possibly noisy geometrical components of an object -- can be suppressed.

The 3-D method is based on [18,29]. The 2-D method is a simple template-checking technique.

input:
An 8-bit binary-valued volume.
output:
An 8-bit binary-valued volume.
parameters:
Default parameters menu is
input from vol # = 0
connectivity = 3-D
# of passes = to completion
shortest skeleton branch length = 1
copy output to vol # = 0
The ``connectivity = '' toggles between ``3-D'' and ``2-D'' -- it determines how the thinning is done.

The ``# of passes = '' can be used to specify the number of iterations of thinning. Leaving the parameter at ``# of passes = to completion'' will thin the volume down to a skeleton.

The ``shortest skeleton branch length = '' parameter specifies the shortest skeletal branches preserved (in voxels).

comments:
 
  1. This can be a slow operation depending on the geometrical complexity of objects in the volume. Thinning a 100x100x100 arterial-tree volume, for example, could require perhaps 20 iterations of thinning and take 10 to 15 minutes.



 

Philip Americus
The Multidimensional Image Processing Lab
Fri Aug 30 10:26:42 EDT 1996