Graylevel Watershed

An ImageJ plugin to apply watershed segmentation on graylevel images

Written by Daniel Sage at the Biomedical Image Group, EPFL, Switzerland

splash image

Outline

This Java plugin is able to segment an image using the watershed algorithm by directly flooding graylevel images. This implementation is in contrast with the classical approach working on the distance map image obtaining after thresholding. The grayscale watershed segmentation is useful to segment particles in contact when the model of shape is unknown a priori.

Reference

[1] D. Sage, M. Unser, Teaching Image-Processing Programming in Java, IEEE Signal Processing Magazine, 2003.
[2] M. Tsukahara et al., Coupled Tomography and Distinct-Element-Method Approach to Exploring the Granular Media Microstructure in a Jamming Hourglass, " Physical Review E, June 2008.

Software

Specification

This plugin works with graylevel images containing bright particles on dark background or dark particles on bright background. It accepts stack of images, but the processing is alwasy done slice-by-slice (2D).

To prevent oversegmentation due to the noise, it is better to denoise first the image with appropriate denoising programs: Median filter or Gaussian blurring are the simple way to do it, more sophisticated filters like Bilateral Filter or edge-perserving filters can give better results. This plugin includes a Gaussian blur filter.

This plugin applies the watershed flooding algorithm which can be interrupted to a user-specified level. Each particle should have a local maximum (or local minimum when the objects are dark) to define a catchment basin. It has various kind of outputs to display the "dams" of the watershed or the "catchment basins" of the watershed. It also includes a fancy colorized animation which helps to understand the flooding process.

Note on the 3D implementation: The watershed alogrithm can be also implemented for isotropic 3D dataset. An example of application for granulometry is shown in the paper of Tsukahara and al. (2009). Nevertheless, the 3D version of the watershed is not provided in this plugin. It is very slow and requires a huge amount of memory.

Pre-requirement

The software provided here is a plugin for ImageJ, a general purpose image-processing and image-analysis package. It runs also on the image-processing package Fiji. ImageJ has a public domain licence; it runs on several plateforms: Unix, Linux, Windows, and Mac OS X. It doesn't take more than a couple of minutes to install.

Distribution for end users

Download Watershed_.jar [Version 05.02.2008]
Put it into the plugins folder of ImageJ or Fiji.

Distribution for macro developpers

Download Watershed-cls.zip [Version 05.02.2008]
Unzip the archive file containing the Java class files and a example of macro.
Put the folder into the plugins directory of ImageJ or Fiji.
To call the watershed routines read the Watershed_Demo.txt.

Conditions of use

You are free to use this software for research purposes, but you should not redistribute it without our consent.
In addition, we expect you to include adequate citations and acknowledgments whenever you present or publish results that are based on it.

Online demonstration

We have also a demonstration of this algorithm running on any browser.

Applications

Examples

Patchwork of results obtain with the plugin

Some citations

Plaque size determination: Stefan A Grathwohl and al. "Formation and maintenance of Alzheimer's disease β-amyloid plaques in the absence of microglia", Nature Neuroscience, 2009.
Comics: C. Ponsard amd V. Fries, "An Accessible Viewer for Digital Comic Books", ICCHP'08 Proceedings of the 11th international conference on Computers, Springer-Verlad, 2008.
Boundaries of paticles in AFM images: N.G. Semaltianos, "Nanoparticle formation by the debris produced by femtosecond laser ablation of silicon in ambient air", Materials Letters 62, 2008.

Screenshot

watershed

daniel.sage@epfl.ch • 06.06.2018