Legacy Documentation

Digital Image Processing (2000)

This is documentation for an obsolete product.
Current products and services
Function Index /
Previous section-----Next section

WatershedTransform

The watershed transform is an image segmentation technique that groups pixels around the local minima of the image and the boundaries of adjacent groupings are located on the ridges separating the minima.
The markers are specified as a list of positions in an n-dimensional array with the n-tuple {xi,0, xi,1,...} denoting the position of the ith marker. For the typical case of 2D images, the markers are pairs of integers, the row and column position of each marker, respectively.
The option ConnectivityNumber may be used to change the default connectivity settings that depend on the depth of the array (see NeighborOffsets for more information). The default setting for 2D arrays is ConnectivityNumber → 8.
WatershedTransform is an implementation of the Vincent and Soille algorithm, extended to arrays of arbitrary depth.
Custom connectivity settings may be defined using a list of offsets.
The argument img may be an array or ImageData expression.
See also User's Guide 6.5.
New in Version 2.

Example

This loads the package.
In[1]:=
This defines an example array.
In[2]:=
Out[2]//MatrixForm=Out[2]//MatrixForm=
Here is the watershed transform of the example array.
In[3]:=
Out[3]//MatrixForm=Out[3]//MatrixForm=
In[4]:=
Out[4]//MatrixForm=Out[4]//MatrixForm=
Users may impose minima on the input array using markers. Here is an example of partitioning the example array into three regions.
In[5]:=
Out[5]//MatrixForm=Out[5]//MatrixForm=