Aralia Systems, a specialist security company, is working with machine vision experts from the University of the West of England (UWE) to redevelop imaging technology currently used for detecting criminals making pirate copies of films with video cameras.
Aralia Systems has been awarded a Knowledge Transfer Partnership with UWE’s Machine Vision Lab (MVL) worth in excess of £215,000 to target the technology for market research data collection techniques.
The project is supported by a grant of £118,000 from the Technology Strategy Board and the EPSRC.
The team will build instruments capable of collecting data that can be used by cinemas to monitor audience reactions to adverts and films. They will also be looking to gather data about attention and audience movement.
According to Prof Melvyn Smith, director of the MVL, the team will combine 2D high resolution camera technology from Aralia Systems with 3D imaging techniques developed by UWE.
Smith explained the 3D imaging is based on a technique called photometric stereo, which was developed as part of a face recognition project at UWE. The technique works by taking three images from the same location but under different lighting conditions.
For the purposes of this application, Smith envisions cinema audiences being illuminated with light in the infrared, a wavelength that is undetected by the naked eye. The light will be positioned in different places allowing the imaging system to recover a 3D map of the face of people sitting in front of the screen.
‘Because our data is richer we think we might be able to acquire additional information about what the audience is doing,’ he said.
‘What we’re keen to determine is: are they sitting in family groups? At what point are they enjoying the film? Is there a point when the section of the audience is looking bored or disinterested?’
Smith said the challenge will be performing this image analysis on the scale of a cinema audience.
‘One possibility is to treat the audience as a kind of texture so you can do simple things like try and determine the direction in which people are looking or maybe isolate individuals which is what Aralia Systems currently do with their 2D cameras,’ he said. ‘Then you can look for more details about facial mannerisms and the mood of the people. Are they looking happy or are they looking bored?’
It is envisioned this will feed powerful marketing data that will inform film directors, advertisers and cinemas with useful data about what audiences enjoy and what adverts capture the most attention. By measuring emotion and movement film companies and cinema advertising agencies can learn more from their audiences that could help inform creativity and strategy.
Aralia and MVL previously collaborated in the area of scene analysis, object recognition and content management, aimed at improving techniques for extracting and differentiating object features from video sequences like CCTV footage.
Aralia has been known more for image processing products in the field of intelligent surveillance and video analytics over the last 15 years, and the company expects its partnership with UWE will help move its technology into the entertainment and marketing industries.
Aralia is currently in partnership with Pikaia Systems, a Canadian company which will be responsible for marketing and deploying the final product. Pikaia holds several pending infrared illumination patents that aid accurate camera detection.