Researchers from the University of York's psychology school have found that faces reflected in your eyes can be captured in high enough resolution to be identified.
To demonstrate not only that the images exist, but that they can be identified, Rob Jenkins and Christie Kerr got volunteers to perform recognition tasks.
Using a 39 megapixel camera, they took photos of their volunteer models, either as the subjects of photos, or as the “bystanders”, appearing only in the corneal reflections of the subjects of the photos.
The corneal images were of comparatively low quality: the “image area for reflected bystander faces was smaller than for subject faces by a factor of around 30,000”, the researchers write.
The researchers applied a bare minimum of post-processing to the photos acting as the sources of the corneal reflections: the images were rescaled to a height of 400 pixels (with bicubic interpolation to reduce noise), and Photoshop's Auto Contrast function was used to improve definition.
The researchers report that their experimental subjects were able to match corneal reflections with the correct person in as many as 71 per cent of cases. In an experiment testing peoples' ability to identify a familiar face in a corneal reflection, recognition rates reached as high as 90 per cent.
The story in your eyes: corneal reflections in (a) yield photo in (c)
The researchers reckon the technique could be useful in investigations of crimes such as hostage taking or child sex abuse, in which the victims are photographed in view of perpetrators.
Jenkins has a long interest in facial recognition. In 2008 while at Glasgow University, he was reported to have developed a 100 per cent accurate facial recognition technique. ®