Watershed


This is a very useful and one of the most used operators of Mathematical Morphology when talking about Morphological Segmentation. The name watershed is due to the basic idea behind the operator, i.e., it finds watersheds. The idea is very simple:

A grayscale image can be thought as a model for a real surface, this you see in the pictures bellow. Simulating a flood in it, the operator can find the dividing lines or watersheds. Figure one is a image of a large square hole with four smaller ones touching it. Figure two is the same image but random colored to better show the depth of each hole. Figure three is the image seen as a surface. Figure four is the result of the operator on the first image. There you see the operator extracted all points where the small squares touch the large one. This is because their borders where at the same level.

fig. 1 - a grayscale image

fig. 2 - The last image with a random color map

fig. 3 - Viewing as a surface

fig. 4 - The watershed

However the operator is not used without some image filtering or preparation because it is very sensitive to noise. If you want to know more about this look in applications.


Building the cantata workspace

Executing the cantata workspace water.wk

Back to the initial page