Whether detecting the faces of terrorists lurking in crowds, locating concealed weapons or identifying explosive devices in subway systems, new 3D technology could be a key weapon in the homeland security arsenal.
Surveillance systems typically rely on two-dimensional images, but if experts can generate high- definition 3D images, security services could harvest much better information.
A team from the University of the West of England’s (UWE) Centre for Machine Vision in Bristol, U.K. aims to extract high-resolution 3D video information even at a distance of several hundred feet.
“This study will test the feasibility of combining photometric stereo with recent developments in sensor and illumination technology,” said Mel Smith, director of UWE's Centre for Machine Vision, in a statement. “We are going to build a demonstrator system able to operate both at close range and at long distance, and will test it in realistic outdoor environments during night and daytime.”
Photometric stereo is a technique for estimating the depth and surface details of an object by examining the same view from different directions.
U.K.-based video surveillance specialist Aralia Systems is also involved in the research project, which is funded by the Technology Strategy Board. The Centre for Machine Vision is part of the Bristol Robotics Laboratory.
If widely deployed, the 3D surveillance technology could help deter terrorist attacks and improve security for civilians.
How does it work?
Essentially, the UWE system analyzes multiple images artificially illuminated from different directions. The technology uses a black silicon sensor and combines cutting-edge detector technology with advanced image processing. Information is extracted and used to build a 3D version of the image, which can be rotated to review from different angles.
The prototype will be tested in outdoor terrain in both light and dark conditions. Airports, train and bus stations, malls, popular tourist attractions, government buildings, ports and more could potentially benefit from the new surveillance technology.
The team’s research also suggests that the technology could work in varying levels of light and dark, as well as obscured conditions involving, for example, fog or smoke.
The technology could also be built into existing security networks such as CCTV.
Using an enhanced remote system could provide advantages like reducing false alarms and the need for a physical security presence.
Detecting and recognizing threats
The Centre for Machine Vision is also investigating how 3D research could help to identify threats like guns or explosives concealed beneath clothes. Scientists have created a portable device that automatically detects and recognizes potential hidden threats to the military in war zones. This, however, could also be useful for protect civilians at home.
The prototype device won funding from the U.K.’s Ministry of Defence under its Competition of Ideas scheme and was developed with U.K.-based systems engineering specialist SEA Group. The technology works by enhancing subtle shapes and surface details, revealing things that are deliberately concealed.
Photometric stereo produces a composite image using light from three or more sources – this is linked to a computer to obtain reference data about a suspected object's surface.
The device is so effective that it can reveal, for example, not just the location of a weapon, but also the type as well. A small wearable version could be deployed on a soldier or first responder to identify threats that are close.
PhotoFace, a facial recognition method that uses high-speed photometric stereo, is another of UWE’s projects. Skin reflectance data is also captured to produce synthetic poses of any face captured by the PhotoFace device.
The PhotoFace project began in 2007 and the initial research phases were in partnership with Imperial College London’s Communications and Signal Processing Group. The team also worked with defense specialist General Dynamics and the Home Office Centre for Applied Science and Technology.
Detecting human emotions
Got a poker tell? UWE’s 4D Vision Project could soon detect harmless poker tells, but also dangerous terrorist ones.
Described by UWE as “next generation photometric technology” the 4D Vision Project captures changing facial expressions in ultra-high resolution. The technology uses a single camera to take a series of 2D facial images while LED light pulses illuminate the face. An algorithm then determines face shape and skin color using shading data. A computer uses this data to create a 3D face moving in real time by processing the next set of 2D images as the 3D face is reconstructed.
Beyond security applications like supporting facial recognition screening at high-risk sites, it could also have potential applications in medicine such as aiding with stroke recovery and facial surgery.