Security cameras able to see through disguises and recognise faces in any light or at any angle
(28/11/2007)

Increased interest in biometric face recognition is timely because of concerns over terrorism. Likely applications include airport immigration control, crowd surveillance, controlling access to secure areas, ATMs and on-line accounts and mugshot identification. Other applications could be in the areas of forensic science, multimedia and communications.

Security cameras able to ‘see’ through disguises and recognise faces in any light or at any angle are a step nearer thanks to new 3D imaging techniques being researched by the University of the West of England and Imperial College London.

UWE and Imperial College are recognised as two of only three UK centres with expertise in the technique, called photometric stereo, which uses subtleties in image shading to isolate and recover fine detail in 3D shapes. UWE and its partners have received £672,000 in funding for the three-year project from the Engineering and Physical Sciences Research Council’s Fighting Crime programme.

Dr Smith said he was delighted to be working with the team from Imperial College, led by Professor Maria Petrou. He added, “As humans we have an amazing capacity to recognise faces. But automated face recognition is one of the most challenging research topics in the field of computer vision. The PhotoFace project aims to develop new forms of capturing 3D images of faces that allow them to be identified despite changes in pose, lighting or facial expression.”

The technique of photometric stereo was first described at the Massachusetts Institute of Technology (MIT) in the 1970s, but remained largely a laboratory curiosity until the early 1990s when a dynamic form was developed at UWE. This allowed moving surfaces with both 2D and 3D features to be analysed for the first time. The team at UWE have also been successful with other bids for funding for potential applications of the technique in fields such as medicine and industry as well as security.

Equipment for gathering data for the project, called PhotoFace, will be set up at UWE’s Machine Vision Laboratory and at the South Wales Offices of General Dynamics UK Limited to collect data for this project.

Dr Melvyn Smith said, “Volunteers can walk through the unmanned device, which automatically detects their presence and scans their face as they walk through. They will be able to see a 3D relief of their face displayed as they exit the system.”

The project is being undertaken in collaboration with Imperial College, General Dynamics UK Limited, the Home Office (Scientific Development Branch) and Identity Solutions Ltd.

UWE’s Machine Vision Laboratory is a member of the Bristol Vision Institute, a group which brings together members from a wide range of disciplines, from biological vision to artificial computer vision systems. The team at UWE consists of Dr Melvyn Smith, Centre Director and Reader in Machine Vision, Dr Lyndon Smith, Reader in Computer Simulation and Machine Vision, Dr Sagar Midha, Visiting Senior Research Fellow, Dr Abdul Farooq, Research Assistant, Dr Gary Atkinson Research Assistant, Dr Jiuai Sun Research Assistant and PhD students Kyle Zhou, Robin Ding and Raphael Wedowski.