Thanks for visiting Imaging and Machine Vision Europe.

You're trying to access an editorial feature that is only available to logged in, registered users of Imaging and Machine Vision Europe. Registering is completely free, so why not sign up with us?

By registering, as well as being able to browse all content on the site without further interruption, you'll also have the option to receive our magazine (multiple times a year) and our email newsletters.

The eyes have it

Share this on social media:

Topic tags: 

Fingerprints and irises provide a unique reference point to verify a person's identity in security and access control applications. Greg Blackman looks at the uses of biometric measurements

The debate still rages in the UK over the effectiveness and necessity of the Home Office’s plans to issue ID cards as part of the national identity scheme. The scheme was set up to provide a secure system of personal identification to UK citizens, and as of 24 November 2008, foreign nationals living in the UK can apply for ID cards at enrolment centres around the country.

Along with standard information, such as name, gender and date of birth, the ID cards will contain biometric data that will be stored on a National Identity Register (NIR). Fingerprints and a face image will both be stored on the NIR and on the chip of the card. Leaving aside the ethical arguments surrounding this topic, the advantage of using biometric data in addition to the standard information is that it confirms not just that you have a valid identity card, but that the card belongs to you.

Essentially, biometrics is the measurement of physiological or behavioural characteristics used to verify a person’s identity. These can include face, fingerprint, and iris recognition, as well as DNA fingerprinting and voice and signature recognition. Automated biometric recognition techniques use complex algorithms to identify unique patterns within images of the distinguishing trait and match them to an individual’s file stored on a database. In this way a person’s eye or finger can act as an ID card, providing them with access to buildings or secure areas.

Fingerprint recognition

Arches, loops and whorls make up the basic pattern of a fingerprint, and their use as a method of identification is well established. Taking an ink imprint of a criminal’s finger is a common procedure for most police departments. ‘Fingerprints were arguably the second major step forward for biometrics, the first being the signature,’ comments Paul Easton, communications director at UKB International, a company providing access security products and services.

UKB International provides fingerprint scanners with multispectral sub-dermal technology. ‘A multispectral imager collects information from below the surface of the skin by reading the sub-dermal collagen ridges that form the fingerprint, in addition to scanning the print itself,’ says Easton.

Easton explains that one of the problems with first generation scanners is that they are dependent on contact of the fingerprint ridges with the sensor. Therefore, anything that interfered with this resulted in an inadequate image. ‘If fingertips were worn, dirty or wet, or the subject did not place their print exactly on the centre of the scanner, the scanner would reject the match. Basically, the door or switch would not work,’ he says.

Human Recognition Systems’ iris recognition equipment has been used in a project to implement biometric methadone dispensers in prisons across England.

Multispectral sub-dermal scanners verify identity even if the fingerprint is abraded, dirty, wet, or covered by a latex work glove. The systems are also much harder to fool. Latex moulds of a valid fingerprint will not activate the switching mechanism. Easton views the technique as highly robust. ‘Although both iris and facial recognition systems are currently commercially deployed, neither is as robust as multispectral fingerprint recognition.’

UKB International supplied its fingerprint scanners to provide gate access at a major US theme park. A temporary fingerprint database was set up whereby visitors’ fingerprints were held on file for the duration of their ticket, and scanners at turnstiles provided them with easy access to rides. Easton notes that it was relatively easy to set up a multi-tiered security system, where customers were given basic park access, whereas maintenance personnel were provided with ‘all area’ access. The scanners provide access to millions of people and are still in operation.

Easton states: ‘Tickets, swipe cards or fobs can all be lost, stolen or “buddied”, that is to say, passed to unauthorised persons. In addition, there is a cost element to providing these tokens.’ Biometrics, however, can provide greater assurance that the individual at the gate is the person who has paid.

Most commercially available fingerprint scanners involve contact of the finger with the scanner plate. The University of Kentucky is currently developing a fingerprint scanning technique using structured light to generate a 3D image of the finger. A bar pattern of light is projected onto the surface of the finger and from the distortion in the pattern, as seen by a camera placed in close proximity to the projector, software is able to generate a 3D image.

Dr Daniel Lau, associate professor in the Department of Electrical and Computer Engineering at the University of Kentucky, explains that contactbased fingerprint scanners present a problem for image analysis software as the finger’s skin undergoes a tremendous amount of distortion from pressing against the scanner. This is especially problematic when rolling the finger across a scanner plate and leads to a drop in recognition performance. The work being carried out at the University of Kentucky uses a noncontact scanner to generate a 3D image of the fingerprint, which can be virtually flattened and analysed and which minimises the problem of image distortion associated with contact scanners.

Providing effective biometric solutions for a specific building or business is relatively easy to achieve, as there are a limited number of employees and the images produced by the scanners will be compatible with those on the database. However, providing the latest fingerprint scanner, such as the one being developed at the University of Kentucky, to national law enforcement bodies like the FBI and expecting it to make matches from an existing fingerprint archive generated from ink imprints is more difficult.

‘One of the great challenges for this kind of technology is making the images compatible with fingerprint data already collected by the various law enforcement agencies,’ says Lau. ‘The problem with using fingerprint scanners is that the image algorithms have to filter the scanned image in such as way as to mimic the ink records.’

Currently, the 3D scanning technology being developed at the University of Kentucky gives a very high recognition rate between scans from the same device, but the percentage drops significantly when it comes to comparing images from a 3D scan with those from a 2D scan. Improving the compatibility with existing fingerprint records is one area on which Lau’s research focuses.

Lau’s team is also using structured light to generate 3D images of the face for facial recognition. ‘An advantage with fingerprint recognition is that it’s a cooperative technique,’ says Lau. ‘The individual places their finger in the scanner to give a reading. In many instances, face recognition is carried out in an uncooperative environment, where changes in lighting and head orientation add to the problems of obtaining a positive match.

‘Three-dimensional techniques overcome many of these issues by being able to subtract out the shadows caused by different lighting conditions and rotate the image to view the head at the correct angle. However, at present, the vast majority of photo ID is two-dimensional and the challenge lies in being able to match 3D images with the 2D records.’

Multispectral sub-dermal scanners verify identity even if the fingerprint is covered by a latex work glove. Image courtesy of UKB International.

Iris recognition

Iris recognition systems utilise the unique patterning on the iris to identify individuals, which is formed from the first year of life and remains stable from then on. Simon Appleton, project manager at the Liverpool company Human Recognition Systems (HRS) in the UK, views iris recognition as being one of the most accurate of all biometric tests. ‘Statistically, it is more accurate than DNA fingerprinting. It’s also non-contact, as the individual simply presents their eye to the device, which avoids problems with hygiene,’ he says.

HRS supplies systems that recognise people and their behaviours, and has been working with the UK Department of Health to implement biometric methadone dispensers in prisons across England. The project aims to ensure inmates receive the prescribed dose of methadone, a drug used to treat recovering heroin addicts.

According to recent figures from the National Offender Management Service (NOMS) responsible for the administration of correctional services in England and Wales, approximately 83,000 offenders are housed in prisons across England and Wales and more than half of these have some form of drug dependency problem.

Methadone is still a dangerous drug and overdoses can damage health and potentially prove fatal. Appleton comments: ‘Manually dispensing the drug is subject to human error and staff can also be fooled by prisoners claiming to be someone else in order to take their prescription.’ The system supplied by HRS uses iris or fingerprint recognition to identify the prisoner and dispense the correct dose according to patient files stored on computer.

Work on iris recognition algorithms was pioneered by John Daugman of the University of Cambridge Computer Laboratory, and many of today’s commercially-available iris recognition equipment are based on this in some form.

‘The prevalence of biometrics will only accelerate as more work is carried out and as the technology matures,’ states Lau of the University of Kentucky. ‘Iris recognition, in particular, is greatly improving. The image quality of high-resolution cameras is increasing to the point where, soon, cameras will be able to capture images of people walking down the street and zoom in on the eye with sufficient image quality as to make an accurate identification.’

HRS is currently developing iris recognition technology to do just that. ‘Current technology is limited to the individual presenting their eyes to the device at a distance of approximately 30cm,’ says Appleton. ‘The product under development is still at the R&D stage, but aims to be able to make a positive identification without the person having to stop moving.’

The use of biometrics as a method of identification is increasing and many companies and organisations are becoming aware of the advantages of the technology. Easton of UKB International comments: ‘Biometric technology has an application anywhere a lock or switch needs to be operated by someone or a group – and not by others. Basic security technology has not changed in centuries. With locks, a bolt is either in or out. A switch is on or off. Biometrics provides a new key, unique to the individual. A key that can never be lost, stolen, buddied, copied or hacked.’

Other tags: