JIRRM is an open source iris recognition software package written in Java. It comes with a single backend library and several front ends, obtainable as a single package or separately.

We are currently working on developing the main library and new ways to integrate the technology with existing software.

Overview of backend design:


Done: The image is cleaned up and a circular Hough transform is used to isolate the inner and outer iris boundaries.
Todo: Advance the algorithm to ensure the inner boundary is within 20-90% of the outer boundary. Automatically choose the iris and pupil boundary radii.

Normalisation and feature encoding:

Done: Basic rubber mapping. Need to account for pupil and iris boundaries not being quite concentric.
Todo: Fully implement rubber sheet mapping - converting the disc of the iris to a rectangular matrix. Remove occlusions - the eyelashes, the eyelids, specular reflections.


Done: Planning.
Todo: Extract an iriscode which maximises differences. Perhaps using Principle Components Analysis, or potentially a choice of several algorithms.

Front ends to the project:

We have a limited SWT frontend. It needs much development, some of which is likely to take place as backend development and testing occurs. There is scope to develop third party plugins, eg. for xlock.