RSS Feeds
Who Are You?
Vendors make it easy to add biometric sensors to security systems
John Titus -- Design News, February 7, 2005
Fingerprints are like opinions; everyone has one—or more. But in the case of fingerprints, they always differ. The use of these unique "personal signatures" has intrigued people since the mid-1800s. Only recently, though, have sensor technologies simplified the task of gathering fingerprint information electronically so people can use it much like a password or PIN. Keep in mind, though, that biometric sensors will always serve within larger security systems that incorporate smart cards, photo-ID badges, RFID tags, or remembered information. Fingerprint sensors will not replace these "keys."
Science fiction stories often describe fingerprint-sensing devices that secure top-secret installations. The technology is now becoming so commonplace that engineers now include it in an array of consumer products. Here's an example: The JumpDrive TouchGuard, a 256-Mbyte USB Flash drive comes with an integrated fingerprint sweep sensor that authenticates a user's identity before it provides access to encrypted files. The $70 unit from Lexar Media matches as many as 10 scanned fingerprints.
New capacitive, thermal, and visual swipe sensors take less space and cost less than area sensors, and each use of a swipe sensor "self-cleans" the unit's window. A representative sample shows the selection from which designers can choose.
Swipe Me
A fingerprint sensor from Fujitsu charges small capacitors and measures the discharge rate caused by fingerprint characteristics at each element. These sensors offer a high signal-to-noise ratio and a resolution of 500 dots per inch (dpi), which, according to Mike Chaudoin, senior marketing manager for Biometrics at Fujitsu, represents the U.S. government's minimum resolution for forensic applications. Each dot produces an 8-bit gray scale value. The direct-contact sensor requires no lens and thus no distortion can occur between a person's fingertip and the sensor.
Chaudoin notes that capacitance sensors work well when engineers need small size, mobility, and low power consumption. "If you plan to monitor a watch list of several-thousand people, a system based on this sensor works well. But if you want to scan a latent fingerprint and match it against prints from millions of people, then another sensor might be more appropriate."
|
| The thermal sensor in Atmel's fingerprint scanner measures temperature differences between ambient conditions and the ridges and valleys on a fingertip. |
Thermal swipe sensors, such as those from Atmel, measure temperature differences between ambient conditions and the ridges and valleys on a fingertip. According to Kevin Heher, business development manager for Biometrics at Atmel, the temperature sensor delivers a perfect image.
"Our sensors can make over 2,000 scans per second," Heher notes. "You can't move fast enough to get a bad reading. A finger moves across a 30- to 50-µm-thick coating on top of the sensor material, so no glass or air separates the finger from the detector." The Atmel sensor produces thermal data that firmware processes to yield a fingerprint image. Key benefits of the Atmel sensor include a thick protective coating that protects the sensor during rugged use and the availablility of flexible all-in-one modules.
The only optical fingerprint-swipe sensor comes from Kinetic Sciences Inc. The patented sensor includes a light source, a prism, a focusing lens, and dual linear arrays of light sensors. "The challenge with swipe sensors involves reconstructing a fingerprint in real time," says Guy Immega, KSI's CEO. "We do it without requiring a position detector."
KSI spaces two parallel arrays of photodetectors in its sensor 150 µm apart. As a fingertip moves across the sensor, the first array puts out its scanned data and a short time later, the same data appears at the output of the second array. "We continuously correlate the two arrays' signals to calculate a finger's speed. Then we use the speed data to remove unneeded scans and construct a complete fingerprint image," Immega explains. "We place the two arrays far enough apart to accurately measure speed yet close enough so the lens simultaneously focuses an image on both." Software can alter contrast and resolution on the fly. The advantages of the KSI sensor include forensic quality images of up to 1,000 dpi, low cost, and a rugged design—finger don't touch silicon.
| Two Ways to Get the Print |
|
| Capacitor-based sensors detect discharge rates that vary due to proximity with fingertip features. The Fujitsu sensor produces scans that a processor turns into complete images. |
|
| Time correlation of the scans from a Kinetic Sciences' visual sensor provides finger-speed information that allows construction of accurate fingerprint data. |
Give It a Whorl
Most security systems don't save complete fingerprint images. Instead, algorithms create a template based on patterns or minutiae in a fingerprint image. Patterns include loops, whorls, and arches, and minutiae cover bifurcations, ends, dots, islands, and enclosures in ridges. The algorithms quantify 40 or so key points in a fingerprint to form a template. In most security systems, rather than trying to match complete fingerprints, software checks for template matches. Storing templates reduces data-storage requirements, speeds matching, and eliminates security issues that surround saving fingerprint images. Given a template, it would prove difficult, if not impossible, for someone to recreate the original fingerprint.
Vendors may suggest that fingerprint technologies offer 100 percent reliability, but they don't. Sensors can have difficulties with children up to about six years old, and finger conditions change from dry to moist, dirty to clean, and so on. Some people's fingerprints have deteriorated due to extreme use. Manual laborers' fingers, for example, may prove impossible to scan because of scarred or calloused fingers.
|
| The fingerprint sensor (gold stripe) in the Lexar USB memory stick scans fingertips and allows access to data only when a match occurs. |
During enrollment, the process of entering fingerprints to build the starting database, an administrator must swipe a finger several times to ensure a good acquisition, and the software must acquire a minimum number of template points to ensure a good match later on. Users should scan several fingers to ensure cuts, stains, or other fingerprint alterations don't block access to a protected area or prevent use of a device. In some cases, a system may reject a person's fingerprints altogether, a condition labeled "failure to enroll."
When evaluating biometric technology, designers should pay attention to manufacturers' specifications such as failure-to-enroll, false-reject, and false-accept rates. A false-reject refuses entry to someone who the system should let through, usually a minor annoyance. On the other hand, vendors don't want their equipment to require such a high security threshold that it routinely blocks a family's access to its home or vehicle.
A false accept, perhaps the most dangerous occurrence, would open a "gate" for someone without proper authorization. But according to several sources, that rarely happens. If a system doesn't see all the necessary template matches, a person doesn't get through. Developers can choose from a variety of "feature-extraction" algorithms and from many template-matching algorithms. The algorithms let developers control characteristics such as the number of template point matches required to get past the security system. A government facility might require a 40-point match, but an office copier might require only a 15-point match. The fewer the matches needed, the faster the matching algorithm can complete its task.
Contributing writer Jon Titus can be reached at jontitus@comcast.net.
| Web Resources | ||
|
Fingerprint sensors and development kits from Fujitsu:
http://rbi.ims.ca/4386-556 |
Kinetic Sciences sensor evaluation kit:
http://rbi.ims.ca/4386-557 |
Thermal swipe sensor evaluation kit from Atmel:
http://rbi.ims.ca/4386-558 |
|
Lexar USB memory stick with fingerprint sensor:
http://rbi.ims.ca/4386-559 |
||
Sensor Drops In, Starts Up
Sensor Drops In, Starts Up
Security systems that employ fingerprint sensors can greatly bolster security. In most cases, however, biometric data will supplement information that individuals carry with them—ID cards and passports—and information they remember, such as PINs and passwords. Datastrip (Shelton, CT; www.datastrip.com), for example, manufactures personal identification and verification equipment that OEMs can customize to meet specific security needs. You might find the company's equipment at airline security counters or at secure government facilities.
|
| A DSVII Biometric Smart Card Reader obtains information from an ID card and matches it against a cardholder's fingerprint. This unit employs an area sensor (above the thumb). |
The company's DSVII-SC Biometric Smart Card Reader, for example, includes a fingerprint sensor. In a typical application, the DSVII-SC reader would accept a smart card that contains an individual's encoded fingerprint information. Then it would attempt to match that data against a "real-time" fingerprint acquired and processed by the reader.
According to Stephen Sardi, senior vice president at Datastrip, the company evaluated about half a dozen fingerprint-sensor technologies over several years to determine which one best met its needs. In the end, the company selected a Fujitsu sensor because of its technical requirements for resolution, cost, ruggedness, and other criteria.
Moreover, because the sensor module incorporates a standard USB connection, engineers quickly connected the module to an embedded computer that runs a version of Windows. The USB interface also simplified the software connection with Datastrip's existing code. "Fujitsu did much of the development work and they worked with us as we integrated their sensor into our product," Sardi says. He emphasizes that the Datastrip engineers took only a few weeks to get up and running with the sensor.
Talkback
DN's Resource Center Get Free Information, Made Easy
-
Quiet Strength Under Load
Groschopp | Case Study
Request Item -
Patient Lift Sound Data
Groschopp | Case Study
Request Item
