Skip to content

Commit

Permalink
update README.md
Browse files Browse the repository at this point in the history
  • Loading branch information
dlegland committed Jul 1, 2021
1 parent b5d24af commit dad7721
Showing 1 changed file with 30 additions and 5 deletions.
35 changes: 30 additions & 5 deletions README.md
Original file line number Diff line number Diff line change
@@ -1,10 +1,35 @@
# Sliding Strel plugin

Implementation of mathematical morphology based filtering using sliding structuring elements,
provided as plugins for the ImageJ software.
provided as plugins for the [ImageJ](http://imagej.net/Welcome) software.
The main purpose of this plugin is to compare running time of morphological operations (dilation, erosion, closing...)
using different implementations of disk or ball structuring elements.

This implementation has the following dependencies:
* ImageJ
* JUnit
* MorphoLibJ (for iamge utilities, and comparison with other implementations)
The development is based on the
[MorphoLibJ](https://github.com/ijpb/MorphoLibJ) library, for comparing with other implementations, and for 3D image management utilities.

## Installation and usage

To use the plugin within ImageJ, simply add the jar file into the "plugins" directory
of the ImageJ/Fiji installation.

You need to restart ImageJ, or use the "refresh menus" command. After that, two new entries appear in the
"Plugins -> MorphoLibJ Plus -> Sliding Strel" menu:

* **Sliding Disk Filtering** allows to perform morphological filtering on 2D grayscale (or binary) images
* **Sliding Disk Filtering 3D** allows to perform morphological filtering on 3D grayscale (or binary) images

Each plugin opens a dialog that allows to choose the type of operation (dilation, erosion, closing...),
the algorithm to use (Sliding, Naive, or ImageJ native), the size of the structuring element, and an option
to display the structuring element.

The "Sliding" version of the structuring element is faster than the naive one. The native ImageJ version uses
multi-threading, resulting in faster running time for "small" structuring elements (usually less than 50 pixels or
voxels, depending on hardware).

## Known bugs or limitations

The definition of the radius is not the same for the native ImageJ implementation as for the two other one.
This results in slight differences between resulting images.

Sliding structuring elements are not multi-threaded, while this is technically feasible.

0 comments on commit dad7721

Please sign in to comment.