Technology

Iris biometric systems rely on detection and recognition of the textured pattern of the iris – the annular region of the eye bounded by the pupil and the sclera (white of the eye) on either side. The boundary between the iris and sclera is often called the “limbic boundary”. The features of the iris are formed randomly during foetal development in the womb and stabilize during the first two years of life (there is evidence to suggest this development is complete within the first few months). They are unique between left and right eyes, and also between identical twins, so are totally unconnected with genetic make-up.

Iris imaging is arguably the least intrusive of the eye related biometrics. Often mistakenly called scanning, it utilises cameras built around common commercial imaging sensors to grab digital photographic images of the iris, and therefore requires no direct physical contact between user and reader. In all current iris image acquisition systems, neither the illumination source nor the imaging plane is scanned across the subject. However, the CMOS or CCD sensors typically used in ALL digital still and video cameras employ internally to the chip one of two pixel information collection techniques known as progressive scan or interlaced scan. Progressive scan sensors are preferred for iris recognition, because there is no need to re-register the alternate rows of pixels after acquisition of successive halves of the image field, as happens with interlaced scan.

For optimum performance across all human subjects, the camera system must illuminate the eye with a low level of Near Infra Red (NIR) light which is needed to reveal the iris texture in eyes that exhibit significant melanin (brown) pigmentation.

Although suggested as an identification technique many years ago, the extraction of features from a digitised image of the iris was originally put forward in a 1986 concept patent by two US ophthalmologists, Leonard Flom and Aran Safir. The first techniques of feature extraction were patented by John G Daugman in 1991, and the method has been demonstrated to work with a high degree of discrimination over a wide variety of ethnic groups.

Since the expiry of the Flom and Safir patent in 2006, there has been a great increase in research and investment into iris recognition systems, Smart Sensors being one of the companies at the forefront of this. Recent hardware and software developments including image analytic processes pioneered by Smart Sensors have led to the development of iris recognition ‘at a distance’ and iris ‘on the move’ generating new interest in the use of iris recognition in situations that require a much lower level of co-operation from the user.

All current iris recognition systems require detection and segmentation of the iris image texture from the digital image of the eye, and then feature extraction which leads to creation of a binary template (also known as a “feature vector”). When an identity determination or verification attempt is made, a probe template is matched against an enrolled template (verification) or a gallery of enrolled templates (identification) usually using a Hamming Distance (HD) score technique which is based on logical exclusive OR. The lower the score, the better is the match between the templates. The characteristics of iris recognition and the HD scoring methods used result in a very high level of confidence in a particular identification result, whether Match or Non-match.

Advantages of iris recognition include:

• Iris patterns possess a high degree of discrimination and randomness in nature; shown to have more than 250 degrees of freedom;
• No physical contact required;
• Protected internal organ; does not “wear” and is less prone to injury;
• Medical evidence shows it is highly stable over the lifetime of an individual;
• Externally visible; patterns available to be imaged from a distance;
• High degree of user acceptance; no criminal connotation;
• Uses common inexpensive digital video sensor technology;
• Much more competitive and economic business landscape.

Research