What is VTBeans?

The VTBeans toolkit was written in Java as part of project FLIERS, to help with visualising and exploring fuzzy classifications of remotely-sensed images. The original toolkit was based on JavaBeans, but the current version is a plain Java application. The toolkit was designed to complement some multi-dimensional VR visualisations generated at JRC, by allowing the user to generate multiple LINKED graphs, animations and 2D surface views. Data brushing techniques ensure that while the individual maps and graphs are simple, they can be cross-referenced interactively to highlight pixels of interest in every open window.

The original VTBeans code was written between 1997 and 1999, and between 1999 and 2004, very few updates have been made (other jobs intervened!). However, I am now in a better position to fix bugs and maintain the code, so bug reports and suggestions would be very welcome, as would any details of contexts where the toolkit has been applied.

Lucy Bastin (04/08/04)

Application version of the VTBeans toolkit

This version of the toolkit runs as a Java application, and can be used with the JDK or JRE. For many purposes, it is therefore more flexible and portable. It also incorporates a number of updates and corrections, and so, unless new users have a particular requirement for the JavaBeans format, they are recommended to download the application version.

To try it out, please download this zip file and follow the instructions.

1. Unzip the file supplied into a new directory (e.g., ' C:\Beans ')

2a - if you have the Java Runtime Environment or Java SDK installed

You can use the 'run.bat' file to start up the application without any edits.

2b - if you don't have the JRE or Java SDK installed

Go to THIS PAGE for some instructions on downloading and installing the JRE

3. Run the application, either with a Windows shortcut to the batch file, or with the command  C:\Beans> run


Using the application toolkit.

On running the application, you will see a small pink canvas appear, with two white buttons. To view the data IO and manipulation options (e.g., file reading & writing, data scaling and manipulation), double click on the pink rectangle and select 'Options' To view the Display options, (e.g., maps and animations), use the 'Display' button. The 'Graph' button will bring up the graphing options described in the tutorial.

  There are several options in this toolkit which are not described in the existing tutorial - for example, a version of Bezdek's fuzzy c-means classifier algorithm. If you would like more information on these or other aspects of the tookit, please contact l.bastin@aston.ac.uk. I would also be very interested to hear about bugs, problems, uses of the tools or suggestions for additions to the toolkit.

Go to the Tutorial!


Updates and bug fixes - 21/1/01

1. Quick viewing of raster data

When a quick-view frame is opened by clicking on a layer's symbol, it;


The layer buttons can still be used for a quick grey-scale view of single layers, but the RGB quick option has been temporarily removed.

 

2.  Interactive scatterplots

The interactive scatterplot now ignores pixels with values of zero or less, making it easier to redraw when datasets are large. The rationale for this change is that when viewing fuzzy memberships, the non-zero memberships only are of interest as far as the coincidence or pulling-apart of cluster centres goes.

3. Data scaling errors

Errors in linear scaling of data layers have been corrected.
 

4. Data normalisation

Data layers can now be normalised to a mean of 0 and standard deviation of 1, using the ‘Normalise layers’ button, which allows any number of the data layers to be selected at one time.
 
 

5. Fuzzy c-means classifier option.

A fuzzy C-means cluster algorithm is available for clustering raster layers or ASCII data files. (Button ‘FCM cluster’). Cluster number, icon or ‘softness’, norm and maximum iterations are all selected using dialogs. For each cluster produced, a new data layer will be appended to the raster data file, with the name ‘FCM0’, ‘FCM1’, etc. This is a direct adaptation of the Bezdek et al. algorithm from the paper referenced below (incorporating their own corrections!). For more details on its operation, please see
Bezdek, J.C., Ehrlich, R. & Full, W. (1984)  FCM: The fuzzy c-means clustering algorithm. Computers and Geoscience, 10: 191-203.
For more information on clustering ASCII data files, please contact l.bastin<at>aston.ac.uk

Updates - 02/08/04

1. Tutorial adapted to describe the application version only, and migrated to Aston University's server. Feedback welcome.