University develops lip-reading technology for surveillance systems

Story link: University develops lip-reading technology for surveillance systems by Jan Harris

Richard Harvey, a senior lecturer in computer vision at the University of East Anglia in Norwich, has secured £391,814 funding from the Engineering and Physical Sciences Research Council, for a three-year project to develop lip-reading technology which could be used in video surveillance systems.

The project will collect lip-reading data which will be used to create systems that can automatically convert lip motions into readable text.

Researchers will develop techniques for recognising head positions, lip shapes, and their related sounds.

Lip-reading technology would help video surveillance systems identify people planning a crime or terror attack by recording the movement of suspects’ lips.

The system would be able to identify key words or sentences which would trigger an alert message to a central console, mobile phone, or other communications device.

Automated lip-reading systems are already in existence, but they are relatively inaccurate and require good lighting and static heads.

The technology being developed by the project can already lip-read between 10 and 30 utterances, with an accuracy of around 50%. With further development the technology could be able to handle natural speech.

The technology has obvious security applications but could also be incorporated in camera phones to allow users to communicate in noisy environments.