Graphic STI
logo EPFL
text EPFL
english only
Biomedical Imaging Group
Daniel Sage
BIG > Daniel Sage > Image Processing > Color Segmentation15 September 2019
CONTENTS

Home Page

News

Events

Seminars

People

Research

Publications

Tutorials and Reviews

Demos

Download Algorithms

Teaching

Student Projects

Intranet

Color Segmentation: ImageJ plugin to cluster color pixel driven by the user input

This ImageJ plugin allows to segment an color image or a stack of color by pixels clustering. The cluster has defined manually by the user through the interface.

Download and Installation

To run the plugin you should first download ImageJ. ImageJ is a free iamge-processing software running on several plateforms: Unix, Linux, Windows, Mac OS 9 and Mac OS X. The whole installation process should not take more than a couple of minutes.

Download the JAR file and put it into the plugins directory of ImageJ:

ColorSegmentation_.jar

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 a citation or acknowledgment whenever you present or publish results that are based on it.

Manual

1. Open a color (RGB) image and select the POINTCROSS tool in the toolbar of Imagej.

2. Choose the first color cluster (e.g. A) and define the color by clicking on the image.

3. Choose the second color cluster (e.g. B) and define the color by clicking on the image.

In the dialog box, you can check the color appareance, and the mean values on R, G, and B channel. You can also control the standard deviation σ, a large value of σ can produce bad result because the color you have chosen if too disparate.

4. Choose another cluster, if you want. The maximum number of cluster is 10.

5. Select the data organization, independant channels or joint channels

6. Select the algorithm, K-Means (faster one) or Hidden Markov Model which add spatial constraints

7. Click on 'Run' to produce an ouptut image

To produce a binary image from the cluster image you may be use the ImageJ plugin Threshold_Colour of G. Landini

Screenshot

watershed