|BIOMETRIC TECHNOLOGY OVERVIEW
The use of fingerprinting is well-known because of its use in forensic science and law enforcement. Large-scale fingerprint technology works by using coordinates of points on the fingerprint where ridges end or split. It is also possible to match the whole fingerprint pattern but such systems are rarely used on a large scale.
There are two main ways of recording fingerprints: rolling and slapping. Rolled fingerprints are used in law enforcement where the maximum print is recorded. Slapped fingerprints only record the pad of the finger but the process is less intrusive. There are different types of fingerprint reader: slap readers (10 prints), single-finger optical readers, single-finger capacitive readers, ultrasound readers and rolled fingerprint readers. It is likely that the ID Cards Programme would use a 10-finger slap reader.
The basic characteristics of fingerprints do not change, although fingerprints can be damaged by injury, burns or wear due to work. When recording fingerprints, it is important the finger is clean because any grease or dirt can distort the image. The image can also be distorted by pressure on the finger that alters the fingerprint pattern.
There are hundreds of fingerprint companies but only four or five provide AFIS (automated finger identification systems). There are several large-scale fingerprint databases including the FBI AFIS database, which has a database of 47 million fingerprints and Ident1 in the UK, which holds six million sets of prints.
Facial recognition works by identifying people according to sections of the face least susceptible to alteration eg. upper outline of eye, sides of mouth, cheekbones.
The two main methods are: local feature analysis and the Eigenface method. Local feature analysis measures the relative distances between landmarks on the face. The Eigenface method looks at the face as a whole and uses combinations of 2D templates that represent distinctive characteristics of a facial image.
Face recognition readers vary greatly in technology and the lighting of the face can have a great effect upon the performance of the technology. There are approximately 10 companies offering 3D technology and less than 100 companies offering 2D solutions.
Iris recognition measures the iris pattern in the coloured part of the eye. Iris patterns are formed randomly at birth and iris patterns are different for every eye. The iris can have more than 250 distinct features compared with 40 or 50 comparison points for fingerprints.
Iris scanning involves a camera capturing an image of one or both eyes. The camera focuses on the eye, locates the iris and accounts for areas obstructed by eyelashes or eyelids. This image is broken into circular grids and each area is analysed for unique patterns. This information is converted into an algorithm in the camera that can be used as a template.
Iris patterns are unique, even between identical twins, and these patterns are stable throughout life. There can be some difficulties with iris scanning if individuals are wearing glasses or contact lenses, if they have aniridia (lack an iris) or glaucoma.
The iris recognition market is currently dominated by Iridian, although it may become increasingly competitive as patents expire. Iris performance statistics from independent tests are limited to 100's. However, the technology is widely used in the United Arab Emirates, which has a database of over 350,000 iris scans.