For scientists and engineers involved with face-recognition technology,the recently released results of the Face Recognition Grand Challenge–more fully, the Face Recognition Vendor Test (FRVT) 2006 and the Iris Challenge Evaluation (ICE) 2006–have been a quiet triumph. Sponsored by the National Institute of Standards and Technology (NIST), the match up of face-recognition algorithms showed that machine recognition of human individuals has improved tenfold since 2002 and a hundredfold since 1995. Indeed, the best face-recognition algorithms now perform more accurately than most humans can manage. Overall, facial-recognition technology is advancing rapidly.

Jonathon Phillips, program manager for the NIST tests and lead author of the agency’s report, says that the intended goal of the Face Recognition Grand Challenge was always an order-of-magnitude improvement in recognition performance over the results from 2002. Phillips believes that the necessary decrease in error rate to achieve that goal was due in large measure to the development of high-resolution still-images and 3-D face-recognition algorithms. “For the FRVT 2006 and the ICE 2006, sets of high-resolution face images, 3-D face scans, and iris images were collected of the same people,” Phillips says. “The FRVT 2006 for the first time measured the performance of six 3-D algorithms on a set of 3-D face scans. The ICE 2006 measured the performance of ten algorithms on a set of iris images. 3-D face recognition has come into its own in the last few years because 3-D sensors for face recognition have become available only recently. What 3-D face recognition contributes is that it directly captures information about the shapes of faces.”

Among other advantages, 3-D facial recognition identifies individuals by exploiting distinctive features of a human face’s surface–for instance, the curves of the eye sockets, nose, and chin, which are where tissue and bone are most apparent and which don’t change over time. Furthermore, Phillips says, “changes in illumination have adversely affected face-recognition performance from still images. But the shape of a face isn’t affected by changes in illumination.” Hence, 3-D face recognition might even be used in near-dark conditions.

According to Ralph Gross, a researcher at the Carnegie Mellon Robotics Institute, in Pittsburgh, 3-D facial recognition can also recognize subjects at different view angles up to 90 degrees–in other words, faces in profile. “Face recognition has been getting pretty good at full frontal faces and 20 degrees off, but as soon as you go towards profile, there’ve been problems.” Gross says that the explanation for face-recognition software’s difficulties with profiles may be no more complicated than the fact that no one was focusing on the problem. The main applications of face recognition have been in contexts like ID cards and face scanners, for which the aim has been recognition of the full frontal faces of cooperative subjects under controlled lighting.