By Erik L. Ridley, AuntMinnie staff writer
November 15, 2010
Thursday, December 2 | 10:50 a.m.-11:00 a.m. | SSQ10-03 | Room S402ABResearchers from Weill Cornell Medical College in New York City examined the potential for facial recognition software to match patient photos with their 3D images, in this study to be presented on Thursday.
Seeking to evaluate the ability of facial recognition software to identify patients based on their 3D surface-reconstructed facial images, the researchers tested Google's Picasa online photo management service, which includes facial recognition capabilities.
In one of three cases, Picasa was able to make the match between 3D facial reconstructions and digital photographs of patients. While in the other cases it couldn't, the software also was never wrong; when it made a match it was always correct, said senior author Krishna Juluru, MD.
As a result, the researchers concluded that facial recognition software has the potential to perform matching between soft-tissue reconstructions of the face and photographs.
"The question that arises is: If commercial software can do it, what can more customized applications do?" Juluru said. "It could potentially have a much better performance."