[ VIGRA Homepage | Function Index | Class Index | Namespaces | File List | Main Page ]
Options object for watershedsRegionGrowing(). More...
#include <vigra/watersheds.hxx>
Public Member Functions | |
WatershedOptions & | biasLabel (unsigned int label, double factor) |
Bias the cost of the specified region by the given factor. | |
WatershedOptions & | completeGrow () |
Perform complete grow. | |
WatershedOptions & | keepContours () |
Keep one-pixel wide contour between regions. | |
WatershedOptions & | seedOptions (SeedOptions const &s) |
Specify seed options. | |
WatershedOptions & | srgType (SRGType type) |
Set SRGType explicitly. | |
WatershedOptions & | stopAtThreshold (double threshold) |
Stop region growing when the boundaryness exceeds the threshold. | |
WatershedOptions & | turboAlgorithm (unsigned int bucket_count=256) |
Use a simpler, but faster region growing algorithm. | |
WatershedOptions () | |
Create options object with default settings. |
Options object for watershedsRegionGrowing().
Usage:
see watershedsRegionGrowing() for detailed examples.
WatershedOptions | ( | ) |
Create options object with default settings.
Defaults are: perform complete grow (all pixels are assigned to regions), use standard algorithm, assume that the destination image already contains region seeds.
WatershedOptions& completeGrow | ( | ) |
Perform complete grow.
That is, all pixels are assigned to regions, without explicit contours in between.
Default: true
WatershedOptions& keepContours | ( | ) |
Keep one-pixel wide contour between regions.
Note that this option is unsupported by the turbo algorithm.
Default: false
WatershedOptions& srgType | ( | SRGType | type | ) |
Set SRGType explicitly.
Default: CompleteGrow
WatershedOptions& stopAtThreshold | ( | double | threshold | ) |
Stop region growing when the boundaryness exceeds the threshold.
This option may be combined with completeGrow() and keepContours().
Default: no early stopping
WatershedOptions& turboAlgorithm | ( | unsigned int | bucket_count = 256 |
) |
Use a simpler, but faster region growing algorithm.
The algorithm internally uses a BucketQueue to determine the processing order of the pixels. This is only useful, when the input boundary indicator image contains integers in the range [0, ..., bucket_count-1]
. Since these boundary indicators are typically represented as UInt8 images, the default bucket_count
is 256.
Default: don't use the turbo algorithm
WatershedOptions& seedOptions | ( | SeedOptions const & | s | ) |
Specify seed options.
In this case, watershedsRegionGrowing() assumes that the destination image does not yet contain seeds. It will therefore call generateWatershedSeeds() and pass on the seed options.
Default: don't compute seeds (i.e. assume that destination image already contains seeds).
WatershedOptions& biasLabel | ( | unsigned int | label, | |
double | factor | |||
) |
Bias the cost of the specified region by the given factor.
In certain applications, one region (typically the background) should be preferred in region growing. This is most easily achieved by adjusting the assignment cost for that region as factor*cost
, with a factor slightly below 1.
Default: don't bias any region.
© Ullrich Köthe (ullrich.koethe@iwr.uni-heidelberg.de) |
html generated using doxygen and Python
|