Fingerprint Sensors

Among the various biometric authentication devices available, the use of fingerprint sensors is expected to spread widely in the future due to their favourable cost performance and sophisticated authentication functions.
Fujitsu offers a range of products which meet the demands of the broadband Internetage, for example the proven FRAM (*1) technology LSI for multi-function smartcards,
and is working actively to create highly reliable security solutions based upon fingerprint sensors (*2).

♠ The background to the spread of fingerprint sensors

The spread of the Internet has made it necessary to verify the identity of individuals online. The simplest form of individual verification is the password, but these do not enable high levels of security. In the United States of America, where the Internet is most developed, statistics show that an average citizen holds eleven passwords. However, passwords that are easy to remember may expose personal property to risk, and it is a fact that net crime is increasing. It is the recognition of this security flaw which has prompted security specialists to develop biometrics. In biometrics, authentication is done using a specific physical characteristic of the individual.
In addition to fingerprints, the voice, iris, hand, face and veins can also be used, but among these, fingerprints are regarded as superior in their cost-effectiveness and authentication accuracy. Capacitive based solid-state sensors achieve stable operation, low power consumption and physical compactness, and as such are surpassing the more conventional optical sensors in the fingerprint sensor market.

♠ Fingerprint detection principle

The surface of the capacitive based solid-state sensors consists of an array of capacitor electrodes (e.g. 76,000 electrodes on the MBF200) covered with a hard protective layer. Varying charges accumulate across this array depending upon the distance between the surface of the finger and each electrode. The sensor reads each capacitor value and performs an 8-bit AD conversion to output an image of the fingerprint. The pitch of the condensers is 50 µm, so the sensor is able to detect the ridges of the fingerprint, which are larger than 200 µm, with high accuracy (500 dpi).

Principle of fingerprint detection

The capacitor charges (C1,C2,C3), corresponding to the ridges and troughs of the fingerprint, are measured.
76,800 pixels of fingerprint information (on MBF200) is converted to 8-bit scale data.

♠ Features of the capacitive based solid-state fingerprint sensors

• A silicon semiconductor device which performs stably with no optical mechanisms.
• The use of standard silicon semiconductor technology means that various types of control circuits, memory, interfaces, etc., can be easily integrated. Proven technology is also used for reduced power consumption.
• Available in various sensor sizes to suit applications.
• Ultra-hard protective coating for durability.

Fingerprint authentication software (*3)

The fingerprint image is read from the chip, and extraction of the fingerprint data, collation and authentication are done by the fingerprint authentication software. There are various different methods for fingerprint authentication, including pattern
matching and minutiae. The latter of these is explained here.
Each human fingerprint has its own unique characteristics.
There are particularly large personal differences in the fingerprint ridge bifurcations and end points, the "minutiae", and these do not change without some physical surgery or injury.  The fingerprint authentication software extracts these minutiae and then discards the original fingerprint image (which is approximately 77 Kbyte).
The fingerprint data registered is a simple template (at most of several 100 bytes), which contains information on the positions of these minutiae.
It is impossible to reconstruct the original fingerprint image from the data stored.
The minutiae are likewise extracted and then compared with the stored template for the authentication of fingerprints.

Principle of Characteristics extraction

• Privacy is protected as the method adopted makes it impossible to reproduce the fingerprint image from the fingerprint data stored.
• A high level of authentication accuracy is achieved as any rotation or
  shift in the positioning of the finger is accounted for.
• Data handling is easy as at most only several hundred bytes of data is
  stored for each fingerprint.

---

*1.FRAM (Ferroelectric Random Access Memory)
A non-volatile memory using ferroelectric material, in which data is stored even when the power is cut off. Has low power consumption and enables high-speed data writing and high re-writes counts.

*2.Fingerprint sensor device
A device which collects a fingerprint image in order to perform personal authentication. Types include optical, pressure, thermal and capacitive. With capacitive sensors it is possible to achieve highly
reliable authentication, high resolution and sensor miniaturization.

*3.Authentication software
Software to perform personal verification by comparing the fingerprint data sensed with data previously registered.