Next: ``Salt_and_Pepper_Removal'' - remove salt
Up: Filter
Previous: ``maximum homogeneity'' - the
- VFX name:
- iSymmetricNN
- IMPROMPTU Equivalent:
- Symmetric_Nearest_Neighbor
- purpose:
- Run the symmetric nearest-neighbor (SNN) mean filter
on an input volume [6]. The SNN mean filter is
a nonlinear, noise-smoothing, edge-preserving
filter. It also can preserve thin lines (depending on the
operator window size) in images (although the sigma filter may
be better for this purpose).
- input:
- An 8-bit volume.
- output:
- An 8-bit volume.
- parameters:
- Default parameters menu is
x_filter_dimension = 3
y_filter_dimension = 3
z_filter_dimension = 3
iterations = 1
The parameters ``x_filter_dimension'', etc., specify the size
of the local operator used by the filter.
- comments:
- 1.
- This filter may ``sharpen'' regions more impressively
than the sigma filter, but the sigma filter will preserve
thin lines better. The maximum-homogeneity filter
sharpens noise and reduces noise better than both the SNN
and sigma filters.
- 2.
- This is a moderate computation algorithm (depending on
the window size).
MultiDimensional Image Processing Lab, Penn State University