Facial recognition, a biometrics approach that mixes computer and security video technologies, has had its ups and downs.

In national legislation, the U.S. Department of Homeland Security for border security fondly embraces it. Then this summer it was kicked out of Tampa, Fla., where law enforcement used it to try to identify the bad guys as they walked the city streets.

Bouncing back from such a setback, facial recognition is slowly but surely being accepted by state and local law enforcement agencies as a unique lineup image enhancement tool.

The makeover of facial recognition depends on specific applications as well as the encouragement of government as federal law and regulations encourage the acceleration of development and implementation of biometrics.

The New York City-based research firm International Biometric Group (IBG), for instance, just released an expanded report “State of Facial Recognition Technology” that shows growth of the sector. It examines significant facial recognition deployments and the impact of facial recognition’s incorporation into machine-readable travel document applications. There also are multimodal biometrics applications, which include facial recognition paired with finger, hand or other biometrics.

The report also examines the emergence of three-dimensional or 3-D facial recognition technology. This newer approach is catching the eye of government, corporate and system integrator buyers.

There’s no doubt, however, that border security legislation enacted in May 2002 by the U.S. Congress has had the most obvious and recent impact on facial recognition.

Growth and Advances

By October 2004, appropriate federal agencies must start issuing biometrics-inclusive visas and travel security documents to foreign visitors. At the same time, nearly 30 countries must add facial recognition into passports since, up until that time, its citizens did not need to have a passport to visit the United States.

The border security applications, in their various forms, do not necessarily watch for the faces of terrorists or international criminals. They instead allow security personnel to more accurately match a live person to his or her facial recognition data housed on a chip on the passport or other ID. The design does not use streaming security video throughout airports, in ports or on city streets but instead uses fixed video at border, transportation and customs checkpoints.

Identifying and capturing criminals on-the-fly using facial recognition and street video, on the other hand, has proven a bust, at least in Tampa, Fla., one of two cities experimenting with the approach.

Tampa police at first installed about three dozen security video cameras in the city’s Ybor City historic district several years ago. The police felt the cameras were helpful in preventing drug and violent incidents. Then in 2001, and thanks to a donation by a facial recognition manufacturer, Tampa police added facial recognition to the video system as well as a face database of about 24,000 known criminals and runaways.

Some Failures

The facial recognition experiment in Tampa ended this summer. Based on results of the experiment as reported in Florida media, the system did not lead to any arrests. There were numerous false matches. On occasion, the system matched a male face with a female identity. And, more telling, attention to the facial recognition portion of the security video system drew police attention away from more routine and effective general street surveillance.

The City of Virginia Beach, Va., is still using facial recognition, although its application is much smaller than the Tampa experiment. Through numerous federal grants, other law enforcement and public safety agencies are using facial recognition at airports, jail visitation centers and courthouses, to name a few sites.

One example is the use of facial recognition technology from Viisage Inc., Littleton, Mass., by the Pinellas County, Fla., Sheriff’s Office. The Pinellas facial recognition project first began with the replacement of the mugshot system used to capture an inmate’s picture and demographic information, including scars, marks and tattoos. Now facial recognition integrates into pre-booking, formal booking and release procedures.

In addition, at select workstations throughout Pinellas County, investigators have the ability to use facial recognition to compare any image in the course of an investigation against the entire sheriff’s office booking database.

Also this summer the Arizona Department of Public Safety upgraded its facial recognition system to now search more than one million facial images. In this application, technology from San Diego-based ImageWare works with a proprietary facial recognition technology from Identix, Minnetonka, Minn.

Law Enforcement, Corrections Applications

“By using the facial recognition system, we have been able to quickly and efficiently streamline our booking and investigative processes, and share that critical information with sheriffs’ offices, corrections facilities and police departments across the state, including agencies such as the FBI, Immigration and Naturalization Service and U.S. Customs,” says Cyndy Pellien of the Arizona Department of Public Safety.

While current technology is two-dimensional, manufacturers and government agencies are introducing or experimenting with 3-D facial recognition.

Viisage and Identix both are working on 3-D facial recognition projects within government-sponsored development programs.

The objective of the projects is to improve the performance of facial recognition technologies when comparing 2-D to 3-D images. This process is the basis for the automatic comparison of photos and live individuals. The application, following the latest recommendations of the International Civil Aviation Organization on biometrics – enhanced passports and border control scenarios, appear to be an important use of facial biometrics into the future.

Another advance centers on the ability of a system to recognize multiple faces simultaneously. For example, Omron Transaction Systems Inc. of Schaumburg, Ill., recently introduced FaceRecoSEARCH, a face identification software library that uses advanced technology to simultaneously recognize multiple faces in a group of people as they approach a monitored area.

“The technology recognizes slight changes in facial images such as rotation, expression and secular distortion,” says Takeki Nemoto of Omron Transactions Systems Inc.

“Using a standard operating environment, identification is typically performed in less than one second, making the technology recognize VIPs as they enter a hotel, identify missing persons or lost children as they approach a public transportation station, for real-time passenger counting or for security monitoring.” Omron also has a facial recognition-based access control system for entry and exit of an access door. Comprised of an access panel containing a charged coupled device or CCD camera and a keypad and a computer control unit, the system stores and calculates individuals’ facial information and monitors how many times a door has been accessed, when and by whom.

SIDEBAR 1: Adding More Dimensions

By deploying a new generation of 3-D facial recognition technology, these variations of illumination and pose no longer would hinder a reliable identification of faces on the basis of photographs. The research project now initiated by the U.S. government is conceived to realize 3-D facial recognition applications and to make available for deployment a highly effective tool that performs a reliable and efficient identity confirmation. The face is represented by an “elastic” grid, which is adjusted to the specific facial pose and expression by adapting the size, position and internal distortion. Within this grid, approximately 2,000 characteristics are used for the facial analysis. The flexibility of the technology and the complexity of analyzed features account not only for superior performance, but also provide an optimal basis for 3-D facial recognition.

SIDEBAR 2: What About the Eyes?

Even before facial recognition, a handful of companies were selling identification and security systems that used iris and retinal scanning as a biometrics.

Advances continue. Companies that are more often identified with security video are also in the iris recognition arena, bundling with biometrics firms.

Covering the recent Government Security Conference, SECURITY magazine discovered that Panasonic Security Systems of Secaucus, N.J., now has an advanced badging system called Panasonic Entry Access Recognition & Logistics System (PEARLS). It allows employees and vendors to register and print temporary identification badges unassisted with detailed entry/egress logs for tracking and report generation. The system employs Panasonic’s recently introduced BM-ET300 and Authenticam Iris Recognition Systems along with a host of related products for turnkey systems operation.

PEARLS is designed for employees looking to enter a facility that do not have proper access identification with them and for visiting vendors with continued access privileges to a facility. The system allows individuals to gain secure access without having to check in with security or personnel resulting in both time and cost savings.

A basic PEARLS configuration consists of a registration station and at least one authentication station, which would be located in a primary lobby or entry point to a facility. Employees and vendors who frequent a facility would first register with security personnel to capture an iris code image and ID photo. Upon entering the facility, individuals then log onto the system utilizing Panasonic’s BM-ET300 iris recognition system for identification and authentication. The logon process takes only a few seconds. Upon verification, a temporary ID badge is generated from a Panasonic color laser printer networked into the system. Multiple lobby authentication stations can be networked for campus-wide applications.