OmniPerception's News & Events

Cargo Security International: Face Facts

01 June 2009

David McIntosh looks at new initiatives in the application of biometrics to transport security.

In the two years since I last wrote about the use of biometrics and intelligent surveillance for this magazine, biometric capabilities in particular have moved on a long way (see Cargo Security International, December/January 2007). Applications that seemed destined to remain science fiction for some years to come are now maturing in practical every-day use. Building on these advances, one very small UK company and one of Britain's largest have teamed up in an ambitious partnership designed to deliver even more.

Investment in innovation

Global defence, aerospace and security company BAE Systems recently announced a new initiative to encourage innovation in, and undertake joint development with, small companies selected for then advanced in-house technology and expertise. Named Investment in Innovation and housed within the company's Strategic Capability Solutions business, the programme targets biometrics, cyber-security, intelligent systems and health management as its initial priorities. With increasing pressure among defence and security customers for improvements in accuracy, reliability and cost-effectiveness, Investment in Innovation is designed to harness small company energy and initiative to help deliver step -changes in performance through new technologies and solutions for customers worldwide.

As part of this new initiative, BAE Systems has teamed up with OmniPerception, a British biometric and computer vision company founded in 2001 and now established as a leading facial recognition supplier. The two companies are working together to develop and deliver the next generation of automatic face recognition products for access control, screening, closed circuit television (CCTV) surveillance and forensic video analysis.

The background to this initiative is an increasingly widespread adoption of facial biometrics in access control and identity management solutions, both in the private and public sectors. HSBC's adoption of OmniPerception's CheckPoint™ facial biometric access control for its UK data centres is a recent example in the commercial world. The decision by Merseyside Police to equip all of their custody suites with Colossus™ automatic face recognition, also from OmniPerception, typifies a similar trend towards facial biometrics in the law enforcement and criminal justice fields. The two partner companies plan to extend these uses to a much wider range of applications in national security, critical infrastructure protection and commercial and personal security - in order to meet customers' increasingly stringent requirements in these areas.

One typical day-to-day identity management tool is, of course, the employee identity card. Typically, this is a swipe card, radio frequency identification (RFID) token or other e-enhancement, combining staff identity documentation (ID) checking with various kinds of automatic e-access control, both physical and logical. These can work well in some environments, but less and less so as organisational size and complexity increases and as security threats rise. This is particularly evident in sea ports and airports, and at land borders. Far too easily lost, stolen, forged or falsified, traditional identity tokens are generally highly vulnerable to misuse. The addition of a personal identification number (PIN) or password can help, but this has other inconvenient side-effects and is, in any event, of very limited additional value. PINs and passwords are notoriously easy to hack and steal. They are also easy to forget, which doesn't enhance either user-convenience or system efficiency. Systems dependent on these traditional approaches are inherently very vulnerable to unauthorised penetration. It is chiefly at the elimination of this vulnerability that the BAE Systems / OmniPerception initiative is aimed. Their plan is to cure the problem by using more secure biometric solutions.

Facial biometrics and cargo security

In the field of cargo security, facial biometrics in general and the OmniPerception / BAE Systems partnership in particular seem likely to have most impact in two main areas: 

• access control, for employees and others (including biometric driver identification/ validation) 
• perimeter and site surveillance (including dock-side, air-side and roadside cameras).

In both areas, it is worth emphasising, as always, that advanced automatic visual intelligence is, and will remain, only a part of the overall security toolkit. Computer intelligence supports, but cannot replace, the essential human vigilance and expertise involved. However, the technological advances that have already been made and those now in the pipeline may be expected to bring significant advantages to security costs and performance, as a support to the human element.

With the growth of widespread concerns about personal and national security, including a demand for better ways to combat identity theft, there is increasing recognition that we need to be clearer and surer about the identity of the people we deal with every day, particularly those seeking entry to precincts and premises.

Staff and contractor access control

In a recent conference address, John Williams, the Head of Group Physical Risk at global banking giant HSBC, gave an interesting example of this. He announced that although his bank had not previously felt the need for any widespread use of biometrics, in 2008 it had decided that for its high-vulnerability areas a new more targeted approach was necessary. In that context, his team had decided that biometrically enabled positive ID was essential. After researching the alternatives and conducting in-house trials of two potential solutions, HSBC chose OmniPerception's CheckPoint™ facial biometric system.

The way the system works is that staff with authorisation to enter must present two proofs of identity. The first proof is either a PIN number or a swipe-card and for the second proof, they simply look into a camera for a moment. Typically, the whole process takes less than two seconds. If the face fits, they gain access. If not, they don't. Among other things, this means of course that stealing a swipecard or memorising someone else's PIN isn't going to do an intruder any good. Without the right face, CheckPoint™ won't open the door.

Such systems work very well within the operational parameters defined for them. However, the recent BAE Systems / OmniPerception announcement is about their plans to extend the operational capabilities of biometric solutions beyond the present boundaries. They list the key elements in the equation as including:

• raising the system's tolerance to variations in pose and expression. This will allow people to approach the camera with less concern about how and where exactly they present themselves (current systems are restricted to 'portrait' style poses: face-on to camera with a neutral expression; future solutions will be more flexible) 
• increasing the speed of operation, without compromising security or user convenience 
• achieving identity checking at greater range. This will extend the distance from the approaching individual that the system's cameras can begin to detect and to recognise them.

Perimeter and site surveillance

While biometric access control solutions can incorporate one or more of a widely accepted range of biometric modalities (facial recognition, fingerprinting, iris scanning, hand geometry and finger vein analysis are common examples), options for long range identity checking are much more limited.

The typical distance between the camera and a person or other item of interest makes it essential that detection and recognition can be done remotely using techniques that do not require contact or close proximity. There are some out-lying biometric candidates for the job (including for example facial thermography and long range voice recognition), but essentially for CCTV applications the only biometric modalities under serious consideration are gait analysis and facial recognition.

Gait analysis (the identification of distinctive traits or characteristics that individuals exhibit in the way they walk) has the disadvantage that very few people have had their gait identified and recorded for comparison purposes. Faces, on the other hand, are common currency. Large facial image databases have existed for many years and -through the issuing of passports, driving licences and other ID systems - they are being constantly up-dated.

Facial biometrics takes the lead

The BAE Systems / OmniPerception development programme, therefore, is concentrating on facial biometrics as the main recognition modality, with a key emphasis on an ever-increasing tolerance to variations in pose and facial expression, as well as immunity to speed of movement and extreme variations in lighting. The starting point is the highly effective current range of OmniPerception identity management tools for access control, law enforcement and criminal justice. We are using these as a base from which to develop wider angle and longer range capabilities for surveillance use - both on a live real time basis and in forensic applications.

At docks, ports, airports and land borders in particular, the need to add a much improved 'computer eye' facility to existing manual systems has long been recognised. Significant progress has been made in this field in recent years, including the work undertaken under the auspices of the I-lids programme at the UK Home Office Scientific Development Branch. Further improvements are in the pipeline. OmniPerception and BAE Systems have announced their intention to be at the forefront of those further developments. Given that no reliable practical solutions yet exist for giving security teams the ability to clearly and accurately identify a suspect in CCTV footage, and given that the demand for such a capability is growing as security concerns rise, we hope that many will be watching the BAE Systems / OmniPerception biometric initiative with great interest.