Over the last decade, there has been a huge growth in the use of biometric screening at international airports and other border ports of entry around the world. Passports now contain microchips that carry personal biometrics for use by facial recognition systems. There are databases that contain tens of millions of records of fingerprints or retina scans. Biometric screening has become the preferred tool in identity management, not just for the Government and its security apparatus but also for business. In the USA, US-VISIT has used biometrics and strategic coalitions to build a system, which assists in securing border ports of entry. The Automated Biometric Identification System, or IDENT as it is known, contains the fingerprints and photographs of more than 143 million people. IDENT uses a biometric system of comparing 10 fingerprints for greater accuracy and the systems have been installed in over 200 US consulates around the world.
Biometrics is all about identity management and comparing a person’s unique traits with a sample template. There are only three unique human traits useful in biometrics: a person’s fingerprints, the iris, and the retina in the eye. Because these traits are unique, a biometric system can confidently identify a person as being the same person whose sample is stored in a database.
Other biometric non-unique traits include facial recognition, which compares certain features of the face such as the distance between eyes, the shape and thickness of the nose, the length of the jaw and the shape of cheekbones. These all seemingly present a distinctive map of a person’s face. Facial recognition though is not unique and therefore is not a positive identification.
Biometrics only works if that person is on record and has enrolled on the system. The system cannot possibly identify someone it has never encountered before. Therefore, the fingerprint, retina scan or facial recognition map must already be on the system as a reference for the system to compare. A comparison against a supplied identity is also quicker. For example with a biometric passport, the identity of the person is verified, so the person’s assumed identity is already known. Identification is on the passport, so the comparison is made against that specific template sample residing on the chip for that person. On the other hand, when dealing with an unidentified person, the system has to check through all the records searching for a match. Even on very fast computers and databases that will take some time.
Fingerprints, despite concern from some experts, have been acceptable as a unique identifier. The ease with which fingers can be sampled during enrollment and then later scanned makes them a good candidate for a biometric sample. Fingerprint sensors work by creating an image of the fingerprint. Police and government require a complete image file when dealing with criminals and the justice system. However, in business, security access systems use a mathematical model calculated on the image, and this is sufficient since it takes up less disk space. Fingerprint biometric technology measures the unique whorls, edges and ridges of the fingerprint. Fingerprint solutions are a considerable investment for businesses, so they are likely to remain the popular choice of biometric analysis, despite iris scanning being more efficient. The US-VISIT system in place at every border port uses fingerprint recognition. The IDINT system uses a ten finger sample to increase reliability and reduce error rates.
The retina scan identifies and stores an image of the blood patterns at the back of the eye. The human retina is made-up of a complex structure of small capillaries that supply the retina with blood. This network of capillaries located at the rear of the eye is unique and not even identical twins share the same capillary pattern since it is not an entirely a genetically determined feature.
A retina scan works by shining a beam of low intensity infrared light onto the back of the eye. Because the blood vessels absorb light differently than its surroundings, they are readily identified. The retina scanner detects the pattern of blood vessels and captures it as an image. The image is then passed through an algorithm, which converts the image into a mathematically deduced code, which is then stored in the database.
Retina scans are quick and very accurate with very low occurrences of false positives and practically 0% false negatives. The drawbacks are that the equipment is expensive and subjects consider it an invasive form of sampling.
Iris recognition is an automated method of capturing the details of the iris of the eye. The iris is a very intricate pattern and unique to each individual. By capturing the pattern via a camera, a computer algorithm can identify the person against a stored reference. Iris recognition matching is extremely fast and accurate with very low error rates. Earlier systems were problematic, because they required adjusting the camera position to match the eye level of each person. However, newer camera technology can capture the image of a subject’s iris without them having to look directly into a fixed camera. The technology can work at ten meters or even when the subject is on the move.
Each human face has distinctive characteristics that identify a person. Humans are very good at facial recognition, much better than computers. Computers scan the face for distinguishable landmarks called nodal points. The human face has over 80 nodal points. Particular areas of interest in biometric screening are:
- the distance between the eyes
- the width of the nose
- depth of the eye sockets
- shape of the cheekbone
- length of the jaw
These nodal points create a face-print, which is a digital representation of the subject’s face and is stored as a mathematical map in the database or on a biometric chip.
As discussed, there are several candidate technologies being tested in the pursuit of more efficient and accurate identity management.