next up previous index
Next: ``Cue-based_Hyst_Threshold'' Hysteresis Thresholding using Up: Segmentation Previous: Segmentation

``Hysteresis_Thresholding'' - thresholding using hysteresis

VFX name:
iHystThreshold
IMPROMPTU Equivalent:
Hysteresis_Thresholding
purpose:
Thresholds a grayscale volume using hysteresis, based on a scheme by Canny. Parameters are set manually by the user.

This function will find all objects satisfying the threshold criteria shown below:
1.
The object must contain all voxels which satisfy the following relationship.
$low\_threshold < v(x,y,z) < high\_threshold$
2.
All voxels within the object must be connected to above selected voxels and should satisfy the following relationship.
$low\_threshold - low\_tolerance < v(x,y,z) < high\_threshold + high\_tolerance$.
3.
The object must be made up of $\ge$ S connected voxels, where S is minimum allowable component size.

Note that $T_1 \geq T_2$ and $T_1, T_2 \geq 2$ for this function.

input:
An 8-bit grayscale volume.
output:
An 8-bit binary-valued volume.
parameters:
Default parameters menu is
lower_threshold = 1
higher_threshold = 128
low_tolerance = 5
high_tolerance = 5
min_comp_size = 10
connectivity = 6
conn_type = 2D
The ``min_comp_size'' parameter sets the value of S described above.
The ``connectivity'' parameter toggles between 6- and 26- connectivity.
The ``conn_type'' option toggles between ``2D'' and ``3D'' and specifies how the operation is done. If 2-D option is selected, then:
a) if 6-connectivity is set, then 4-connected 2D analysis is done.
b) if 26-connectivity is set, then 8-connected 2D analysis is done.
comments:
See below.
1.
This function is comparable in run time to the 2-D conn-comp and 3-D conn-comp operations.
2.
See Cue-based_Hyst_Threshold for a comparable function that uses an input seed region to select parameters.


next up previous index
Next: ``Cue-based_Hyst_Threshold'' Hysteresis Thresholding using Up: Segmentation Previous: Segmentation
MultiDimensional Image Processing Lab, Penn State University