In the five years since 9/11, airport security has undergone a paradigm shift. Instead of collecting data following an incident or targeting one of many individual security threats, the focus is now on pre-empting situations. To this end, two technological breakthroughs -- intelligent video and face recognition -- are changing the way law enforcement handles airport surveillance and security.
Adm. David M. Stone is an expert on airport security. He is the former assistant secretary of homeland security for the Transportation Security Administration and served from 2002 to 2003 as the TSA's first federal security director at Los Angeles International Airport (LAX), one of the largest and busiest airports in the world. He currently serves on the advisory board to Vidient.
To explain the paradigm shift in security, Stone likes to use the analogy of an archer. "If you have an enemy archer that has 10 terrorist arrows in his or her quiver, we used to be focused on how you defend on each of those individuals arrows," he explained. "Now we're focused on going after the archer before he or she launches those arrows, which is a much smarter way of doing business."
Smarter because, according to Stone, going after each of the "arrows" reduces the overall probability of finding and preventing security threats. "You're not going to be able to effectively detect and deter each one of those," he said. "Your chances are far better to go after the individual archer."
Intelligent Video
This proactive approach is reflected in the increased use of advanced video technologies in airports throughout the country. Think about a typical airport in the United States. After checking in at the ticket counter, customers make their way to the security check.
As passengers walk over, they spot several security workers. Those workers are stationed at a few key places -- the top of the down escalator to baggage claim, in front of a "personnel only" door, next to the exit of Terminal B. When they reach the security check, they see there's a long in front of them. While there are several metal detectors available, only a few are open, operated by a handful of security staff. The passengers ahead start grousing, frustrated at the wait.
Maybe intelligent video could help.
Brooks McChesney is the CEO of Vidient, which produces SmartCatch, an intelligent video system. SmartCatch works in conjunction with existing airport CCTV systems and, as McChesney explained, is "able to detect aberrant or suspicious behavior and ... distinguish those patterns of activity from normal behaviors. The behavior-based software notes an anomaly and sends a video clip to a responder to act in real time via pager, laptop, cell phone, control center, etc."
***image1***
It's even persistent: "If a responder does not take action," continued McChesney, "the video clip is sent to the next available responder in a specified period of time, such as a few minutes until someone takes action."
Sounds good, but why should an airport consider integrating intelligent video into their existing security setup? "If you went to Los Angeles International Airport today, the largest origin and destination airport in the world, you [would see] an older video system in place," said Stone. "You have various cameras stationed around the nine different terminals; they're fed into a recording device. Not all those camera feeds are continuously monitored. There may be a panel in a corner in a room where you may have one person with a cup of coffee and a clipboard doing multiple tasks but not really watching each one of those monitors. So if you had an event at LAX, the ability in a real-time manner to see something developing is very difficult."
Not only is it difficult, but it's extremely costly. As Stone explained, if a security officer stationed at an exit doorway spots a person -- and maybe it's just a confused commuter -- bypassing security and walking into the terminal, that security officer would call into central command with a physical description and the entire terminal would have to be frozen. Flights would be delayed, passengers would be frustrated, fuel would be burned, all while the video system was accessed. It might take another 15 minutes to review the footage, and then send the information to the terminal so security can look for that individual.
"Most of our video systems at airports have that sort of system in place," said Stone. "It doesn't have the capability to give you a real-time cueing of an anomaly and, because you don't have the money in most places to hire 100 people to sit and watch 100 monitors full time, your ability to have situational awareness is not really there. You don't really have a clear picture of what's happening."
Anomaly Analysis
The San Diego County Regional Airport is not one of those places. Through an informal beta agreement, they are working with Vidient to develop and test an algorithm to detect anomalies at entryways leading to controlled airport security areas.
"Specifically, we are testing this specially developed 'video analytic' at several high-profile access points that transition employees from our terminal public areas to our aircraft ramp and baggage makeup areas," said Mark Denari, director, aviation security and public safety, San Diego County Regional Airport Authority. "Once active, we'll be able to send anomaly data back to our command and control [C2] center for analysis, assessment, and/or response."
***image2***
San Diego International Airport's reason for pursuing intelligent video is its desire to continually improve security operations. "With the current security system so heavily dependent on human involvement and performance," said Denari, "it's extremely difficult to detect anomalies -- suspicious activities, behavior, and/or persons -- given the fact that most of those participating in the security program are engaged in air commerce functions. Although most airports nationwide have law enforcement, security, and air operations personnel assigned primary security responsibilities, these resources are limited and simply aren't enough to achieve true detection capabilities. Hence, the use of intelligent video.
"Airport-centric video analytics," she continued, "especially those developed to specific conditional and operational criteria, can be a powerful multiplier in establishing capabilities that truly detect anomalies."
Those anomalies could be anything, especially since the use of intelligent video isn't just confined to airports. But the programming behind intelligent video can be customized to meet each customer's needs (and anomalies).
Stone explained the flexibility of Vidient's technology: "[Vidient] finds out what, from a security point of view, the leaders of that airport would like to be cued of in real time so that they can respond immediately. The software engineers can tailor software for all of those cameras and then feed your installation, your command center ... This ability lets you to tailor the software to whatever the facility is, whether it's the Sears Tower, the New York Metro, LAX. Each of them are different, they all have things they would like to be notified immediately about, and intelligent video allows you to have a real-time prevention and deterrence capability."
Not only can the software be adapted, but its use can relieve security officers of mind-numbing jobs. "You want people at the screening stations, not at exit lanes or staring at cameras," said Stone.
Intelligent video can monitor traffic lanes, restricted entries, escalators, and the like, letting trained personnel do what they're best at: interact with other people. "Intelligent video frees them up to use their individual talents and relieves them of the monotonous duties of staring at an exit lane or at a camera monitors," said Stone.
Face Recognition
Another technology at work in airports, though not as widespread as some critics would like, is face recognition. Currently, the United States government is conducting the Face Recognition Vendor Test 2006, the latest in a series of large-scale, independent evaluations for face recognition systems.
The FRVT 2006 is measuring the progress of "prototype systems/algorithms and commercial face recognition systems" since the previous test in 2002. It also plans to evaluate performance on high-resolution still imagery (5 to 6 megapixels), 3-D facial scans, multi-sample still facial imagery, and pre-processing algorithms that compensate for pose and illumination.
How does face recognition work? "Each algorithm developer has its own set of features. At one end of the spectrum tends to be pattern matching; it looks at the pixel values that go in and tries to match the pixel values with a new face," explained Dr. Jonathon Phillips, program manager, FRVT 2006. "But it's not just pixel matching -- they do different types of representation. For example, emphasizing the difference among faces but also removing variations between lightings and smiles."
If that sounds a bit complicated, perhaps you'll understand why face recognition isn't more prevalent. There is a large difference between matching up two pictures of well-lit faces looking toward the camera and matching one of those photos with the another person in a different pose, under different lighting conditions, and with a different photo resolution.
"Face recognition is viewed on a spectrum, where different applications fall under different levels of maturity based on the application," said Dr. Mohamed Lazzouni, CTO of Viisage, which produces FaceFINDER. "For instance, matching images in large databases is a very mature business. There are countries that manage attempted identity theft and fraud in their national ID and passport databases by using face recognition technology.
"Similarly, in the U.S., many states use face recognition technology to mitigate identity theft and fraud in their driver's licensing systems. These databases range from hundreds of thousands to millions, and face recognition technology is able to quickly and accurately find potential matches in these databases."
Face recognition technology also works well in the area of verification for access control, and there is continued work on improving this one-to-one match. However, according to Lazzouni, "The area where the technology has not yet matured is in the area of surveillance. Contrary to the portrayal of face recognition technology and popular culture, the technology cannot easily pick a face out of a crowd."
Access Control
Because of this major limitation, face recognition can typically be used in two ways in an airport setting. The first is in access control. In areas restricted to employees or vendors where a badge is used for access, face recognition provides a comparison against a person's badge image, ensuring only authorized individuals gain access to restricted areas.
It's also used in watch list screening. A live image of passengers can be scanned against a database of known criminals to identify those attempting to gain unauthorized access or trying do harm.
The Berlin International Airports use FacePASS to control access to restricted areas. Prior to installing face recognition, employees and third-party contractors allowed access to these areas would present a BIA-issued ID, which would be manually verified by a security official stationed at the entry point. After this proved to be too time consuming, BIA switched to FacePASS.
The process is now automated, verifying each person's identity before they are allowed access. Unauthorized individuals who attempt to gain access are logged and photographed so they may be later be investigated.
Of the installation, Rainer Borchardt, head of airport security, said it was "very positively received by all parties involved -- employee representative, airport staff, and security experts. Automatic face recognition enjoys a high user acceptance, a face which helps safeguard smooth and reliable system operation."
Another advantage of using face recognition is that an organization may already have begun to compile a database without realizing it. "Generally people volunteer to have their photo taken for either a passport, visa, driver's license or other ID," said Lazzouni. "This makes up the vast majority of photo databases."
Lazzouni also noted that using images for identification purposes is familiar to most people. "We intuitively recognize one another by faces and images," he said. He further points out that the technology is non-invasive, meaning that no physical extraction is required. "Also, since the infrastructure already exists, face recognition is a cost effective solution for identity documents and identifying people's records in databases. Finally, the accuracy of the technology is continually improving, particularly in controlled environments."
It's important to remember that face recognition technology is currently best used in controlled settings. Along with intelligent video, as the technology continues to evolve, its uses should become broader and, hopefully, airport security will continue to improve.
MORE INFO
Face Recognition Vendor Test frvt.org
Viisage viisage.com
Vidient vidient.com
|