I have a constantly updated presentation about »The future of computing«. One chapter of it is about security and surveillance technology – the face recognition approach in particular. Two computer science students in Haifa, Israel, have invented a face recognition method with a 3D scan. It can radically improve the success rate and it was even able to seperate them apart: they are twins. The problem is that this approach requires a database of 3D scans of the faces to be matched with the current sample: So how would you collect those 3D datasets?
Now there is a research group in United Kingdom showing a 40 millisecond 3D scan from a single video frame (see demo video). It uses a projection of stripes on the face and then calculates the surface from their distortion. This could even become a way to recognize faces from a running video… and combined with light sources/camera sensors outside of the visible wave length I suspect it could even be possible to aquire the 3D scans without notice of the scanned person. I don’t know the approach is capable of recognizing people wearing glasses or a large beard.
[found via omnio.org]