In the actual face matching experiment, two groups of volunteers were asked to make a judgment. The “Unfamiliar” group was comprised of 16 undergraduate students drawn from non-Psychology departments who may not have previously encountered any of the subjects. The "familiar" group was also comprised of 16 undergrads, but these individuals were pulled from the same department as the subjects. Image pairs of the eight subjects were presented to the volunteers in a random order, with instructions to pair the photo on the left (poor quality) with the photo on the right (good quality), with the ultimate goal of determining whether each pair is the same person, or two different people.
Any faces that were unknown to an observer in the familiar group or known to an observer in the unfamiliar group were excluded from analysis, in order to prevent the data from being compromised. Accuracy in the unfamiliar group was at 71% while accuracy for the familiar group came in higher, at 84%. In addition, another test of spontaneous recognition was performed; where observers could name a face from an eye reflection image they were shown. Accuracy for the correct naming of a familiar face was at 90%, while the mistaken identification of unfamiliar faces was only at 10%.
While this method does require high resolution cameras, the research presented in the paper suggests that corneal image analysis could in fact be useful for identifying perpetrators in crimes in which victims are photographed.
View the research paper.
Also check out:
MIT develops hand-held scanner for detecting eye disease
Facial recognition software used to make arrest
Five machine vision applications to keep an eye on in 2014
Share your vision-related news by contacting James Carroll, Senior Web Editor, Vision Systems Design
To receive news like this in your inbox, click here.
Join our LinkedIn group | Like us on Facebook | Follow us on Twitter | Check us out on Google +
Page 1 | Page 2
