Hendon Publishing - Article Archive Details
Biometrics is Scientific Help
Written by Albert Varga
The technology of facial recognition is either here now or coming soon to your sphere of law enforcement. This aspect of biometrics is not new. Great Britain was one of the pioneers in the use of the technology, and now many other nations, INTERPOL, and the FBI are involved.
Terrorists attacks on our homeland in 2001 created a need for tighter security and increased law enforcement. This was followed by greater interest in biometric technology, especially facial recognition. Today, many county and city police agencies use facial recognition systems.
Fictional television cops use biometrics consistently, and within one hour, they have a suspect arrested. They can pick out the suspect photographed on crowded streets of Manhattan and make a positive identification. Is facial recognition that good in real life? The answer is…yes and no.
“We are proud to be pioneers in the use of facial recognition,” said Sheriff Jim Coats of the Pinellas County, FL Sheriff’s Department. One of the more progressive police leaders, he says, “You can teach an old dog new tricks.” Coats, a 37-year veteran, oversees 3,200 employees and serves more than 300,000 citizens. The sheriff deems his facial recognition program a success.
His deputies can make an identification of a suspicious person on the street using the digital camera installed in the police car. As the suspect stands in front of the patrol car, he is photographed. The photo is entered into the patrol computer, linked to a database accessing nearly 5 million photographs.
The software will take the suspect’s photo and make a comparison. Matches are sent to the officer via his cell phone and computer with physical descriptions of the possible matches. The officer is sent at least three varied photos of a possible match. If it is a match on a wanted person, the deputy makes the arrest.
Captain Jim Main of the Pinellas County Sheriff’s Support Services oversees the facial recognition system. “The photographs of an arrested person or county inmate are entered into a database of photos programmed into digital data and algorithms to establish a positive identification. Using this system, we have made over 400 arrests by mobile identification since it began in 2002.”
He said the system establishes and requires 100 points of identification of the face. Main stressed, “For the facial recognition system to work, you need good clear photographs. I can’t emphasize that enough.”
This facial recognition system, a product of L1 Identity Solutions, Stamford, CT, creates a database that can be shared with any other police agency. The system also includes photos of suspected terrorists. The original cost of the system was about $5 million, and the annual maintenance costs are about $150,000.
Contrast this success story with the Tampa, FL Police Department in 2001. Officials there attempted a different method of facial recognition by installing police cameras in their busiest zone. The system was to identify perpetrators and reduce crime. Few photos were clear enough to make arrests. One alarm from the system reported a suspect on one of the streets. When officers responded, they found the system had photographed a face on a T-shirt worn by an innocent citizen. By 2003, the system was removed.
A similar story occurred in Virginia Beach, VA in 2003. Police installed stationary cameras in a busy zone in hopes of identifying people from the large pedestrian crowd and compare them with a wanted person database. The system became faulty, caused frequent false alarms and became a distraction. By 2006, the system was dismantled.
Both failed systems involved photographing unwary citizens on the move or in day-to-day activities. This type of facial recognition system is improving, but it is the most difficult for law enforcement to employ.
Harris County, TX Sheriff’s Department equips its officers with mobile biometric testing devices, which take fingerprints and photographs of the suspect, which are linked via software to a database. In 2006, officers using such a device arrested a murder suspect in a cold case when a filed comparison of fingerprints and photographs proved positive.
Palm Beach County, FL has invested in $1.1 million for palm print technology to assist law enforcement officers working across municipal boundaries to have access for positive identification of suspects.
In December 2007, the FBI invested nearly $1 billion to create a database of the many fields of biometrics. This national step will help computerize biometrics to identify people in the United States and Canada. This database is available to police, corrections, Homeland Security and other government agencies, including the military. Certain private enterprises such as security firms can use the data in background investigations and security checks.
The benefits for law enforcement are overwhelming. Hardware is available that enables portable technology checks for fingerprints, iris checks, facial recognition and other fields of biometrics. Officers can use cell phones to transmit and receive biometric data. Cold case investigators throughout the country are reporting the benefits of biometrics in clearing cases and arresting suspects. All systems require the police agencies to have the necessary software and the hardware to link to this vast source of data.
For a better understanding of facial recognition, we consulted with Barry Hodge, CEO; Steve Rehfeldt, lead programmer; and Nick Abaid, executive vice president of SecurLinx, a biometric company in Morgantown, WA. SecurLinx works with the University of West Virginia and the West Virginia High Tech Consortium (WVHTC) in developing biometric applications, primarily facial recognition and fingerprints. This was the company that provided facial recognition surveillance to the 2001 Super Bowl in Tampa and the 2002 Winter Olympics in West Valley, UT.
According to his firm, “Biometrics is the measurement of physical characteristics of individuals that make them unique. In law enforcement, these unique characteristics are used to identify someone, the most widely used biometric being fingerprints.”
Abaid is a retired FBI agent. He spoke about the Super Bowl operation. “The Tampa Police had primary responsibility for the Super Bowl and worked with us to install a facial recognition system as an enhancement to their existing video surveillance system. In some ways, we were flying by the seat of our pants since this was the first time facial recognition was deployed at a major sporting event.”
The Super Bowl had a command center to monitor all video screens. If a match was made with the data bank, the information and image of the person was sent wirelessly to the police officers’ portable units in the stadium. The officer or officers could then survey the suspected person closer or stop the person. The data bank consisted of suspected pickpockets, scalpers and possible terrorist.
“Taking the knowledge developed at the Super Bowl, especially the importance of both capturing the facial image and then matching it to the database to get a quick “Yes,” “No,” or “Maybe” reply, we developed a system whereby we combined the best characteristics of facial recognition products of two well-known biometric vendors for capture and matching in a more and efficient manner,” Abaid said.
It is clear that many biometric vendors work together in creating solutions. “There are certain thresholds an analyst uses to measure or define the face to a match. Usually there are a certain number of points to look for, but sometimes it is still up to the police officer or security person using the film to say, that is the guy!” As a law enforcement tool, facial recognition for surveillance purposes is the most difficult application because of the variables involved, such as lighting, camera angles, shadows, the wearing of hats and glasses by subjects and facial angles. However, for secure entry, facial recognition provides an efficient and reliable alternative to cards or passwords, which can be lost, stolen or misused. This system is also valuable tool in identifying people who may not have fingerprints on file or attempt a false identity.
In a police patrol application, the officer receives facial recognition proof or fingerprint verification from a national or state data bank on a portable unit, possibly a cell phone. The officer can document recognition or compare it with photograph identification such as a driver’s license and compare it with the data bank.
Similar to other biometric vendors, SecurLinx works with a police agency or security client to license its application develop software for the biometric application. This includes biometric software applications for opening door locks or surveillance camera work.
Along with facial recognition, there are many other fields of biometrics including palm prints and structure; fingerprints, especially with AFIS; retina scan, which is used primarily for security; iris scan, which is used primarily for security; and voice recognition.
Other less well-known biometrics include ear lobe configuration; handwriting; keystroke, i.e., how one uses the keyboard; and gait, which is detected through an observed way of running or walking. Other biometric technologies as complex as brain scans for lie detection are under development.
Computers have hastened the development of biometrics using algorithms. A layman’s definition of an algorithm is a numerical relation between collections of lists of a quantity (as in known facial features of a suspect); matched with a variable (arrest photo of a possible suspect). The comparison is made in a step-by-step mathematical process to make a final calculation to determine if you have a hit.
When a police agency, enters the market for biometrics, for security or criminal detection, it should seek professional assistance. As officials select a biometric application vendor, they should ask for a formal proposal. This should contain a list of references, a list of all plans, how they will implement the plans, hardware required, software required, training required, total costs, service and back-up.
They should also be willing to work with other vendors. Frequently one vendor is a specialist in fingerprint technology, and they will work with specialist in facial recognition of other forms of biometrics. The proposal should be in laymen’s terms with enough detail to allow an agency to understand what is provided.
For police administrators planning to use facial recognition or other biometrics, consider some of the benefits. Facial recognition programs such as Amber View, developed by the WVHT and the Justice Department, enroll children into a database using school photographs. The parents are asked to pay an additional nominal fee to cover the cost of entering the photo into the national data bank. This improves the safety and chance of identifying children who are lost, kidnapped, or murdered.
Patrol officers would have enhanced portable equipment to make on-the-street identification of suspected persons or fugitives. Security of public facilities will be improved using sure. Identification of fugitives and suspects will be accurate and faster. Increasing databases are available to all agencies. Uniting other biometrics together increases the accuracy of the identification.
In facial recognition, safeguard is a paramount word. Access to this data must be protected. According to the FBI, audit trails are maintained on the users of the data, and there is a periodic audit of each agency that uses the database.
Albert J. Varga is a retired deputy chief of police in Hamilton, NJ and police director in Lambertville, NJ. He is currently a senior manager at Jersey Professional Management, Cranford, NJ, a consulting firm for government services. He can be reached at firstname.lastname@example.org.
Published in Law and Order, Jun 2009
Rating : Not Yet Rated
Related CompaniesL-1 Identity Solutions SecurLinx
Related ProductsBiometricsFacial Recognition Technology
Click to enlarge images.