We noticed you're browsing in private or incognito mode.

To continue reading this article, please exit incognito mode or log in.

Not a subscriber? Subscribe now for unlimited access to online articles.

Intelligent Machines

Boosting Biometrics

Merging multiple biometric techniques creates identification systems that are virtually spoof-proof.

The demand for greater security at borders, government buildings, and companies has meant boom times for biometrics-technologies that measure biological traits to identify individuals. Systems that digitally fingerprint people, read the patterns of their irises, measure the unique dimensions of their faces, or verify their voices are expected to become a $1 billion business in 2004, quadruple what it was just five years ago.

But there’s a problem: no single measurement works for everyone. As many as 3 percent of people lack readable fingerprints, and perhaps 7 percent have eye pigmentation that interferes with iris scans. Face recognition software can be thwarted by veils or thrown off by changes in hairstyle or lighting. And biometrics can be tricked: a fingerprint left on a sensor can potentially be lifted and used by someone else; many face recognition systems can be fooled by photographs or video clips. “No biometric has proven to be the ultimate,” says Gary Strong, the manager of behavioral and biometrics programs in the U.S. Department of Homeland Security’s science and technology office.

Now, corporate and academic labs worldwide are tackling these weaknesses by merging multiple biometrics into systems that are flexible, accurate, and virtually spoof-proof. These new, so-called multimodal biometrics generally take a probability score from each biometric measurement and combine them to provide a single thumbs-up or thumbs-down. Given the pressing demand for better security, revenue from multimodal biometrics is expected to soar from $11 million in 2003 to $220 million by 2008, says Trevor Prout, marketing director of the International Biometric Group, a biometrics consultancy in New York, NY.

This story is part of our June 2004 Issue
See the rest of the issue

Software made by HumanScan of Erlangen, Germany, uses face recognition, voiceprints, and lip motion to identify people-the first commercial multimodal biometric identification product. It starts by preparing a data template for each person who might later need to be screened. A standard video camera equipped with a microphone records one second of video and voice, and the software uses that data to create a unique template. Later, the template can be used to verify identity based on all three signatures. Managing director Robert Frischholz says HumanScan’s combination yields far higher accuracies than individual biometrics. “If you add them all together, you get better results-much better results,” he says.

HumanScan’s technology is already being used to protect certain restricted military computer networks and to safeguard casino customers’ money from being claimed by imposters. But this year the company, in collaboration with IBCOL, a technology commercialization company based in Munich, Germany, is moving to pilot installations that will verify the identities of travelers entering and exiting the United States and Germany. The company will not discuss which biometrics are being used. But there’s already a growing lode of biometric data on travelers. For example, the United States requires that all visitors (except those from Mexico and Canada) submit to digital fingerprinting and photography.

Meanwhile, biometrics leader Identix, of Minnetonka, MN, is enhancing its two existing systems-face recognition software and fingerprint scanners-by adding a third biometric to the mix: skin texture. Identix’s new identifier is called a skinprint, and it’s captured by algorithms that extract texture patterns from digital camera images. “Dermal textures are unique, and they are random,” says Joseph Atick, president and CEO of Identix. Atick says adding texture recognition to Identix’s existing face recognition systems-an advance the company is bringing to market this year-boosts accuracy from 70 percent to more than 90 percent. Atick says that with the new technology, face recognition systems can distinguish between identical twins, making them as accurate as fingerprinting-long considered the gold standard of biometrics.

The texture technology can take advantage of an enormous base of existing face recognition systems, a base that only promises to get larger. For instance, the United Nations’ International Civil Aviation Organization has recommended that passports and other travel documents include chips that carry face data for identity verification, with either fingerprint or iris data permitted as secondary biometrics. Each country would collect this data during the passport or visa application process; biometric systems at airports and other entry points would verify holders’ identities by scanning faces, fingers, or irises and comparing the data against the template on the microchip.

Identix is also experiencing new demand for its software, first released two years ago, that can combine any two biometrics. The technology, known as fusion software, essentially merges confidence scores from the two tests to verify a person’s identity. In one of the more high-stakes examples, Israeli authorities are using the software to merge face recognition with a less frequently used biometric-hand geometry-to monitor the entrance and exit of 50,000 workers to and from the Gaza Strip.

Expanding on this combinational approach is a goal of Homeland Security research efforts. Agency-sponsored research has already created a laboratory prototype that joins three, four, or even more biometrics. The brainchild of Anil Jain, an electrical engineer researching biometrics at Michigan State University, the system can mix and match any combination of face recognition, fingerprinting, iris scans, hand geometry, and voiceprinting. What’s more, Jain says, the system can give greater weights to specific biometrics depending on the reliability of different modes for a given individual.

Some issues must be addressed before multimodal biometrics dominate the industry. Standards must be set, governments need to clarify which combinations will be accepted, and researchers need to determine how best to weight individual scores. But fusing biometrics is critical to truly reliable identification systems. “There is no such thing as one biometric wins all,” Atick says.

(Erlangen, Germany)
Analyzes face, voice, and lip movements to secure physical and network access
(Minnetonka, MN)
Fuses data on skin texture with fingerprint or facial data
Vijayakumar Bhagavatula, Carnegie Mellon University
(Pittsburgh, PA)
Integrates data from face, fingerprint, and iris biometrics
Mike Fairhurst, University of Kent (Kent, England) and Neusciences (Southampton, England) Software agents to manage multiple biometrics, including voice, face, and fingerprint
Anil Jain, Michigan State University
(East Lansing, MI)
Fuses data from different biometrics, including face, fingerprint, iris, hand geometry, and voice, and from different companies’ systems

Learn from the humans leading the way in intelligent machines at EmTech Next. Register Today!
June 11-12, 2019
Cambridge, MA

Register now
More from Intelligent Machines

Artificial intelligence and robots are transforming how we work and live.

Want more award-winning journalism? Subscribe to All Access Digital.
  • All Access Digital {! insider.prices.digital !}*

    {! insider.display.menuOptionsLabel !}

    The digital magazine, plus unlimited site access, our online archive, and The Download delivered to your email in-box each weekday.

    See details+

    12-month subscription

    Unlimited access to all our daily online news and feature stories

    Digital magazine (6 bi-monthly issues)

    Access to entire PDF magazine archive dating back to 1899

    The Download: newsletter delivery each weekday to your inbox

You've read of three free articles this month. for unlimited online access. You've read of three free articles this month. for unlimited online access. This is your last free article this month. for unlimited online access. You've read all your free articles this month. for unlimited online access. You've read of three free articles this month. for more, or for unlimited online access. for two more free articles, or for unlimited online access.