Main Page   Groups   Namespace List   Class Hierarchy   Alphabetical List   Compound List   File List   Namespace Members   Compound Members   File Members   Concepts

WatershedSegmenter Class Reference
[Watershed-based Segmentation Filters]

#include <itkWatershedSegmenter.h>


Detailed Description

This filter implements the first step in the N-d watershed segmentation algorithm. It produces a segmented, labeled image from a scalar-valued image input. This filter is used in conjunction with WatershedSegmentTreeGenerator and WatershedRelabeler to produce a final segmented image volume. See itk::WatershedImageFilter for an overview of the entire algorithm and notes on the terminology used in describing it.

The filter is designed to operate in streaming or non-streaming mode. For more information, see the itk::WatershedImageFilter documentation.
Input
There is one input to this algorithm, a real-valued (scalar) itk::Image of arbitrary dimension. The input is assumed to represents a height function, such as a gradient magnitude edge image. The filter can process an image of any dimension. Note that the terms ``pixel'' and ``voxel'' are interchangeable in this and other watershed component class documentation.
Outputs
There are three potential outputs of this algorithm described below.
The first output is a labeled image of unsigned long integers. This is an initial segmentation and labeling that is fed into successive components of the watershed algorithm.
The second output is a table of segment information, itk::watershed::SegmentTable. This table is a record of each segment numbered in the initial segmentation (output number one) with relevant information needed in successive stages of the algorithm.
The third output is a data structure containing boundary pixel information, itk::watershed::Boundary. This data is only generated if the flag DoBoundaryAnalysis is set to true and is only useful in streaming applications.
Parameters
Threshold is specified as a percentage (0.0 - 1.0) of the maximum height of the image. This filter thresholds the input image to remove all values below $ L = min + T * (max - min) $, where $ max, min $ are the maximum, minimum values in the image and $ T $ is the threshold parameter value. Values in the image less than $ L $ are raised to $ L $.
Thresholding minimum values in the image decreases the number of local minima in the image and produces an initial segmentation with fewer segments. The assumption is that the ``shallow'' regions that this thresholding eliminates are generally not of interest.
See also:
WatershedImageFilter


The documentation for this class was generated from the following file:
Generated at Mon Mar 12 03:34:56 2007 for ITK by doxygen 1.5.1 written by Dimitri van Heesch, © 1997-2000