This demonstration shows the segmentation of images by the watershed algorithm.
The implementation is based on the immersion process analogy, using the gray levels as a topological relief.
The objects to be separated should be dark, i.e. low gray values.
This demo provided a pre-processing filter (Gaussian filter with σ as tuning parameter) to denoise the image first in order to avoid over segmentation.
A Java version is also available as an ImageJ plugin.
 D. Sage, M. Unser, Teaching Image-Processing Programming in Java,
IEEE Signal Processing Magazine, 2003.
 M. Tsukahara et al., Coupled Tomography and Distinct-Element-Method Approach to Exploring the Granular ...,
Physical Review E, June 2008.
Your browser does not support the HTML5 canvas element. Please, try with a newer browser.