Fingerprints: biometrics at your fingertips
A fingerprint contains so-called “level one” (or “coarse”) macroscopic information, consisting of the general direction of grooves. These patterns allow an initial classification of fingerprints into several families (tented arch, left loop, right loop, etc.).
Under a magnifying glass, a fingerprint reveals “zebra stripe” patterns. This examination shows what specialists call “minutiae points”, which is level-two information. There are two types of characteristics: either a ridge ending or a ridge bifurcation, where a line separates. These points make it possible to distinguish between two individuals. One of the primary tasks of a minutiae-based automatic fingerprint recognition system is to locate all the minutiae points in the fingerprint. The system identifies as many points as possible and associates each point with spatial and angular coordinates.
To illustrate this process, the system at the “Biometrics: the body as ID” exhibition represents the minutiae points using a circle with a small tail that indicates the angle of the point. The system memorizes only this set of minutiae points in the template used for subsequent comparison. The next time an individual places his finger on the sensor, the system compares the minutiae points captured with those already recorded. If the two sets have enough matching points, the system confirms that it is the same person.
This is the way the basic system functions. In practice, this type of biometric system also performs a host of other useful tasks. To begin with, it employs a variety of techniques to make sure that the finger presented is “real” to counter fraud, for instance using a plastic finger. The software can also compare the stored print with the newly acquired print, despite the fact that people will hardly ever place their finger on a sensor in exactly the same way twice in succession, with the orientation of the print and the captured zone of the finger varying according to how the finger is positioned on the sensor. It also takes into account variations linked to the elasticity of skin (a fingerprint is distorted depending on the pressure exerted when placing the finger against the sensor). Beyond these “core” techniques, each manufacturer of course has his own particular recipe: “The basic principles behind fingerprint recognition are well-known and documented, but the precise methods used in systems are industrial secrets,” explains Paul Welti, Biometrics Project Manager at Sagem Défense Sécurité.
The principal advantages of fingerprint-based ID systems are accuracy, efficiency, ease of use and implementation and cost-efficiency. On the other hand, these systems suffer from a negative “police state” image, particularly in France. There is also resistance, albeit less widespread, due to hygiene concerns, a somewhat exaggerated fear considering that a fingerprint sensor is neither dirtier nor cleaner than a door handle.