Progress In Unconstrained Ear Recognition (Science 2.0)
The Southampton team have developed a new technique using scale-invariant feature transform and homographies calculated from SIFT point matches. It can cope not only with fuzzy and degraded images, but also with ears that are up to 18% occluded.
A healthy degree of skepticism is in order regarding the short and medium term prospects for the adoption of more exotic biometric modalities: ear, gait, gluteus maximus, etc. In order to make it into commercial applications (as opposed to forensic applications) either ear biometrics are going to have to accomplish an ID task more accurately and more cheaply than the more established modalities (face, finger, etc. which have a head start) or they’re going to have to facilitate identifications that the others can’t.
Maybe ear biometric verification on a mobile phone’s front-facing camera can work.