User Tools

Site Tools


3D Filters

3D filtering using ellipsoidal neighbourhood.


Thomas Boudier

Matthias Labschütz contributed a fast version for isotropic kernels (for min, max and mean filtering)


The plugin performs 3D filtering on image stacks. The kernel is ellipsoidal, you can then select three radii for X, Y, and Z. Note that the plugin uses parallel implementation, and a very fast version using rolling ball algorithm is used for isotropic kernel (all three radii equal).

The 3D filters mean, median, variance, min and max are implemented in ImageJ since version 1.47.

Available filters :

* Mean and variance filtering

* Minimum and maximum filtering

* Local maxima

* TopHat transform (to detect spots-like objects)

* Sobel (simple edge detection)

* Adaptive Nagao-like

3D variance


For details go to the 3D ImageJ Suite.

If you want fast version for non-isotropic kernels, you are invited to install the JNI version.


When using the 3D Filters plugin for publication, please refer to :

J. Ollion, J. Cochennec, F. Loll, C. Escudé, T. Boudier. (2013) TANGO: A Generic Tool for High-throughput 3D Image Analysis for Studying Nuclear Organization. Bioinformatics 2013 Jul 15;29(14):1840-1. doi


GPL distribution (see license).


  • 22/06/2011 : Parallelization updated
  • 22/10/2012 : bug fixed for filters3D with 32-bits images
  • 28/05/2013 : new filters Open and Close
  • 10/04/2017 : new timer for computing time

Known Bugs

  • ij1.47d : Bug in Color/Split, the resulting stacks have incorrect bitdepth
plugin/filter/3d_filters/start.txt · Last modified: 2019/04/12 13:13 by

Donate Powered by PHP Valid HTML5 Valid CSS Driven by DokuWiki