Electronic Design

Biometrics Looks To Solve Identity Crisis

New technologies will use fingerprint, iris, facial, and even vein recognition to provide identification—but at what cost to privacy?

You see them in blockbuster movies and high-tech TV shows—biometric systems that rely on fingerprints, facial recognition, and other physical and behavioral data to provide identification. But these technologies have moved past the sci-fi genre, and even beyond the high-security arena. They’re hitting the mainstream now. In fact, you may even be using some of them already.

Of course, companies in this segment are working hard to keep one step ahead of their competition and criminals alike. Faster and more accurate technologies are arriving, often created by merging multiple sources like fingerprint, iris, hand-geometry, foot, voice, RFID, and vein recognition data.

By going “mainstream,” it’s no surprise forecasts call for a strong ramp up in the sector. According to the International Biometric Group’s Biometrics Market and Industry Report, 2007-2012, revenues will grow from $3 billion last year to nearly $7.5 billion by 2012 (Fig. 1).

Human Recognition Systems, a multibiometric systems integrator in the U.K., is testing out a system for multi-modal biometrics. Sponsored by Manchester Airport and the U.K. Department of Transport, the Bio-Sec trial system is assessing the practical and user acceptance levels of multi-modal biometrics in an airport environment. Technologies on trial include iris and hand recognition working with photo ID systems. Meanwhile, the Japanese Ministry of Land, Infrastructure and Transport is working with the JAL Group to test fingerprint- and facial-recognition multimode systems at Tokyo International Airport in Narita.

TRIED AND TRUE FINGERPRINTING
Fingerprints are the best-known and oldest form of biometrics. Commonly available, biometric fingerprint sensors use capacitive technology to verify users and guard against the unauthorized use and theft of such electronic items as laptop computers and mobile phones. These sensors also can be found in keylessentry automotive systems.

AuthenTec’s AES2810 low-power fingerprint authentication sensor suits notebook computers. According to the company, it’s the first single-chip sensor of its type to integrate a proprietary RF-based sensor, a hardware security module, and a matching engine that performs 128-bit encryption and decryption. Operation is based on AuthenTec’s TruePrint subsurface fingerprint technology, which can read fingerprint patterns from anyone under a wide variety of conditions (Fig. 2).

Upek, Inc. has capitalized on a number of key SmartFinger patents from Norway-based Idex covering ac capacitive fingerprinting sensing. Behind this technology, the company’s TouchChip product family uses a one-dimensional stripe geometry (Fig. 3). A number of other chip manufacturers make fingerprint sensors, too.

Fujitsu integrates a touch sensor onto its FOMA F905i mobile phone. Atmel offers the AT77C102B thermal fingerprint sensor. Infineon Technologies AG makes the FingerTIP capacitive fingerprint sensor and the SICRYPT secure token platform, which was implemented in the Smart Card funded by the European Commission. And, the PFC2020 fingerprint biometric processor ASIC from Fingerprint Cards AB in Sweden acts as a data-processing subsystem for the company’s FPC 1011C sensor and links to the sensor and to external flash memory for storing fingerprint templates.

Fingerprint identification sees widespread use despite the fact that it’s a slow process—it requires an average of 5 to 10 minutes to “roll” a single fingerprint. It also is subject to potential sources of errors. Typical records are taken by pressing the finger or fingers against a solid sheet of paper or a pad, but the pressure can vary, and details can be warped or smudged.

Also, contamination is possible if the paper or pad has been used already. Cuts and callouses can compromise fingerprint identification. Simple inattention to procedure, as fingers must be completely rolled from side to side during the process, can jeopardize integrity as well. Researchers at Warwick University in the U.K. are working on a system that can identify partial, scratched, smudged, or otherwise warped fingerprints in just a few seconds.

Nonetheless, fingerprints are still an effective ID method. The U.S. Federal Bureau of Investigation uses the Integrated Automated Fingerprint Identification System as a database for criminal apprehension and enforcement. Also, the U.S. National Institute of Standards and Technology (NIST) will issue its fingerprint-based Personal Identification Verification (PIV) smart cards to all federal employees and contractors seeking entrance to federal facilities (Fig. 4).

READING AT A DISTANCE
Driven by the federal government’s need to rapidly, accurately, and more efficiently scan fingerprints, the U.S. National Institute of Justice has already submitted applications to fingerprint system developers for its Fast Fingerprint Capture Program. It’s calling upon machinevision technology as a solution for non-contact fingerprinting, with funding provided by the U.S. Department of Homeland Security.

Machine-vision systems already inspect items on production lines and conduct crowd surveillance. Researchers believe that combining these capabilities with other biometric modalities like facial recognition could lead to accurate remote, non-contact fingerprint reading.

Northrup Grumman hopes to have a prototype remote fingerprint system ready later this year. Using standard megapixel cameras, the system would scan the subject’s fingerprint from a range of 1 to 2 meters, in addition to the subject’s iris or face. So far, researchers have used the Bozorth3 algorithm developed by NIST to generate an image, including 52 minutiae scanned by their system, that’s comparable to an image taken from a standard ink print.

Continue on Page 2

Similar efforts at the University of Kentucky and commercial partner FlashCan 3D are developing systems that scan a hand to generate a 3D image in real time and convert that image to simulate a 2D rolled fingerprint. The technology uses structured striped lighting where illumination is projected on a fingerprint in a pattern.

“This type of lighting allows the collection of extra information, such as the depth of fingerprint ridges,” says associate professor of electrical and computer engineering Daniel L. Lau. “It is important to understand that for a remote fingerprint system to be practical, it must make use of relatively low-cost consumer cameras that can work at low levels of lighting.”

The relatively higher cost of a remote fingerprint system is a factor, considering the camera, the lighting source, and other hardware and software. But so is the relatively large size of such a system compared to a conventional fingerprinting approach. Nevertheless, for cases where higher levels of accuracy are required such as homeland security and highly classified access to government facilities, remote fingerprint readers can prove useful.

Iris-recognition systems use pictures of the iris, which is unique, stable, and reliable. These flexible, non-contact, non-invasive systems also offer speed and unmatched accuracy compared to other security alternatives at distances of 3 to 10 in.

Recent developments have advanced facial-recognition systems, too. For instance, the Face LogOn Xpress software from XID Technologies visually recognizes computer users when they log on. Enabled by the company’s facial-synthesis and recognition technology, it works with readily available low-cost Web cameras.

Used in surveillance applications, Cognitec’s FaceVACS technology accurately recognizes people regardless of their facial expression, age, or other variables like hairstyle, glasses, or lighting changes. Also, Oki Electric Industry Co. developed image-processing hardware IP to enable face-detection functions in an IC. Known as the Face Sensing Engine (FSE), this face-processing middleware targets high-end mobile phones and digital cameras.

Recognition Systems markets the HandKey and HandPunch access control products (now re-branded as Schlage HandKey and HandPuch products). These products are based on an image-acquisition system originally developed at Michigan State University to test imaging’s usefulness in hand-recognition biometrics. The system comprises a light source, a camera, a single mirror, and flat surface with five pegs on it (Fig. 5).

After capturing the image of the hand, the system extracts key features for authentication and identification, such as the widths and lengths of a finger at various locations. Users place their hand palm down on the flat surface. The five pegs serve as control points for the users’ right hand. Controls are available to change the lightsource output intensity and the camera’s focal length. The mirror projects the side view of the hand onto the camera. All of this information is fed into a computer for analysis. Current research involves identifying new features that would permit better discrimination between different hands as well as deformable models.

HARDWARE/SOFTWARE DEVELOPMENT TOOLS
One measure of interest in biometrics can be seen in the large number of hardware- and software-development tools on the market. As part of National Instruments’ LabVIEW platform, the BiometricsVIEW CM fingerprint scanner toolkit lets developers easily integrate Verifier 300 fingerprint scanners, which come by way of Cross Match Technologies, to their applications.

The toolkit also works with the Crypto-G comprehensive cryptographic library from Vartor Technology Solutions. Meanwhile, Lithuania-based Neurotechnology offers the Verifinger 6.0 softwaredevelopment kit for fingerprint recognition and the VeriLook 3.2 software-development kit for facial recognition.

A MATTER OF PRIVACY
So what are the social ramifications of these new security technologies? It depends on who you ask. Some people welcome improved methods that safeguard and verify their identities. On the other hand, not everyone is willing to provide such private data, fearing it may get into the wrong hands and be misused. The dangers of identity theft have increased in the 21st century on all levels, from personal finances to national security.

Still, progress can be seen in the biometrics purview. Two years ago, several biometric techniques were applied to the new passport introduced by a dozen nations from the European Union. This e-passport uses a digital photograph of the bearer taken to exact specifications for machine facial recognition.

Continue on Page 3

Additionally, the passport includes an RFID chip that contains the bearer’s biometric data, which is encrypted to prevent identity theft, though authorized computers can read this data at a range of about 50 feet. These types of biometric chips are beginning to appear in airline boarding passes as well.

The EU’s goal is to ensure that highly accurate biometric systems can indeed work while allaying privacy fears. Though advances continue apace in all of these biometric modalities, the jury is still out on how rapidly and widely the application of the technology can be used, since there will always be doubters.

Critics point out that hackers have successfully compromised the e-passports, often as the result of software coding errors in the passport reader machine. Machine reader manufacturers, however, have stepped up efforts to safeguard against these unwarranted intrusions by more rigorously testing their software and by providing effective measures like “Faraday” shields.

Recent biometrics research is now looking at vein recognition, which captures a vein pattern via infrared light. Deoxidized hemoglobin, a blood component, absorbs the light and causes the veins to appear as black patterns that are then translated into a mathematical representation or a template.

Hitachi developed a grip-type fingervein authentication technology for door handles, enabling secure access control. Techsphere also offers hand vascular pattern recognition biometrics for secure access control. Fujitsu came up with a contactless palm-vein recognition unit that recognizes unique palm vein and contour images (Fig. 6). And, the Vascular Pattern Scanner from Identica scans a person’s veins below the skin on the back of the hand.

Ultimately, DNA can be the one “true” biometric identifier. In fact, strides are being made in DNA lab-ona- chip systems. But DNA analysis is a slow process, and it is ill-suited for present-day authentication and identification purposes like airport screening, financial transactions, and access control. However, DNA identification may not be all that far away.

TAGS: Automotive
Hide comments

Comments

  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
Publish