ImageJ Plugins

I have been involved in image processing for more than thirty years now and in this time have accumulated a large number of image processing algorithms which I feel may be of use or interest to a much wider community, especially now that sophisticated image processing can be done on your average laptop or desktop PC.

For this reason I have decided to convert the algorithms to ImageJ plugins, and these will appear on this page over time. Source code and precompiled class files are included. Documentation is normally in pretty basic HTML which means that you can use them easily without needing to be an expert in Javadoc or needing to have proprietary word processing packages installed.

This collection is a work in progress as I centralise material from various places and I'll post a note to the ImageJ discussion forum whenever anything on this page is updated.

I've chosen ImageJ as a vehicle for three reasons - (1) the tool itself is excellent and powerful, (2) it's free, so available equally to all, and (3) it's written in Java so should run on any platform. Of course, there is a trade-off involved in the use of java, which boils down to sluggish performance and high memory consumption. Routines that have high computation requirements will therefore run much faster in a language like 'C' (or even -  gasp! - FORTRAN) and in these cases I may well include 'C' implementations as well for those who are able to exploit them.

ImageJ itself can be found here, and I highly recommend it.

I've placed all these items under the GNU 'Copyleft' umbrella. Inn short, you are free to use them without restriction for personal and research use, but commercial use requires my permission (which I wont withhold unreasonably!)

Filters

• A fairly substantial collection of convolutional filters can be found here. They are collected into groups of related items but all have the following in common:
  • they all use simple matrix math only so you will not find filters that require other operations such as sorting (e.g. median and modal filters) in this group (they will eventually appear elsewhere)
  • they all have a convolution kernel that defines the filter Note that over the years, the same kernel (i.e., filter) has sometimes appeared in the literature with more than one name. You will find some of these duplicates in this package as I opted for completeness and clarity.
  • they are designed to work on images (as opposed to classified output maps, for example)
  • the implementations are all for 8 bit unsigned integer data - either greyscale or the intensity channel of a colour image