Every advance in the ability to recognize faces in digital images adds to the saying, “You can run, but you cannot hide.” This applies not only to criminals, terrorists, and local thugs but also to political dissidents, profiled minorities, and business competitors – it all depends on the circumstances. The technical circumstances of facial recognition have always been difficult. Put a face in a studio with optimal lighting and the camera/software combination will get recognition right almost every time. Put the same face in a brightly lit street full of motion and shadows, and the recognition rate goes toward nil.
The CheckPoint.S system developed by the firm Omniperception (Surrey, UK) reckons on natural lighting being forever difficult to use for facial recognition. Instead, the new system uses its own lighting – near-infrared light emitted from the camera. Infrared light is invisible to the human eye and is somewhat less complicated to analyze (limited bandwidth) for specific biometric features.
Of course, it does require specialized software to use the infrared images. (Infrared images can be made visible to humans. The outlines of things are normal, but the lack of coloration, something like a black and white negative image, is strange to the eye.) The CheckPoint.S system has new algorithms to pick out facial features in their infrared reflection, and convert to those features to a biometric system that can be compared to a database of faces. Other software, developed by BAE Systems (UK) helps analyze face images that are moving or taken from an unusual angle.
The combination of controlled lighting with motion and angle correction helps make the facial recognition good enough to provide alerts in real time. It is not, yet, good enough for legal recognition. It could not be used to identify someone as being in a particular location in a court of law. This sort of precision is still the domain of closed circuit television (CCTV) and human eyeballs. What the CheckPoint.S system does do is provide an automated and almost immediate ‘first alert’ so that other systems (that is, people) can complete the recognition.
‘The thing that has plagued facial recognition for many years is the effect of lighting,’ OmniPerception’s CEO Stewart Hefferman told The Engineer. ‘Our systems operate in the near-infrared space rather than visual light.
The government indicated in its coalition agreement that it intends to introduce legislation to regulate CCTV. Hefferman said he would welcome regulation as a way of addressing public concerns about security systems and invasion of privacy.
[Source: The Engineer]
OmniPerception’s CheckPoint.S system was announced September 1, 2010 and will be immediately available.