Atmel solves fingerprint riddle

Capturing fingerprint images has become one of the most common methods of biometric identification. It's fast, accurate and easy to use. And fingerprint scanners can be integrated into many types of devices — from notebooks to smart-card readers to automated teller machines.

The problem is, as with many other technologies, there are trade-offs. Most fingerprint scanning technologies are difficult to integrate into small devices and require frequent cleaning, but Atmel Corp. has come up with a new solution called FingerChip that promises to solve those problems.

A Little Background

Fingerprints consist of ridges and valleys that contain features called minutiae. Examples of minutiae include Y-shaped junctions and places where ridges end. A fingerprint generally contains 30 to 40 minutiae. Typical fingerprint recognition algorithms convert the minutiae into a template that is saved and encrypted. For security, the fingerprint image itself is never saved.

Fingerprint images can be captured using one of several types of technology, including optical, silicon or thermal.

The oldest method is optical scanning, which is widely used. Most optical scanners use a Charge Coupled Device (CCD) — the same type of device used in digital cameras — to capture the image of a fingertip that is placed on an illuminated plastic or glass platen. The CCD then converts the image into a digital signal.

Optical fingerprint scanners are reliable and inexpensive, but because they are fairly large, they cannot easily be integrated into small devices. And latent prints left on the platen by previous users are a persistent problem because they interfere with accuracy.

Silicon-based scanners use capacitance, or an object's ability to hold an electric charge, to capture fingerprint images. Charge-sensitive pixels in the device distinguish between ridges and valleys, each of which carries a different electric charge. These scanners are small enough to integrate into devices such as notebooks, and the image quality is generally better than that of optical scanners.

Thermal fingerprint recognition employs the same technology used in infrared cameras to measure the differences in temperature between the ridges and the air caught in the valleys. When a finger is initially placed on a thermal scanner, the temperature difference between it and the pixel array in the sensor is large and therefore measurable, so an image is created. But the image disappears quickly; it takes less than one-tenth of a second for the finger and the pixels to reach an equal temperature.

This limitation spurred Atmel to create a unique, patented method of capturing fingerprints called sweeping. Instead of placing and holding the fingertip on a capture window, as is required with the methods outlined above, the fingertip is swept vertically over a small rectangular window that scans it in sections.

The window, or "image zone," is about the width of a fingerprint but only 0.4 millimeters high. Because of its small size, the chip is inexpensive, costing only $5 to $6 each for high-volume production.

The scanning process creates a sequence of horizontal image frames that are reconstructed into one fingerprint image using Atmel's proprietary software.

The scanner is small enough to be integrated into small devices such as personal digital assistants (PDAs). In fact, Hewlett-Packard Co.'s Compaq iPaq h5400 Pocket PC features an Atmel FingerChip.

Atmel provided us with a sample scanner and demonstration software so we could experience the scanning process. We found it easy to capture a good image on the first try, and the process is extremely fast. All it takes is one quick sweeping motion of the fingertip. Users must be sure to sweep the entire fingertip and apply a small amount of pressure while sweeping.

We were surprised, however, to find that if we held a fingertip on the scanner, it quickly became very hot, to the point of discomfort.

The ultimate test of the FingerChip's success will be time, but considering its small size and low cost, don't be surprised if it starts to make more frequent appearances, especially in small devices such as PDAs and wireless phones.