3D Face Recognition

   Ron Kimmel
   CS Dept. Technion

At the Geometric Image Processing Lab., together with Alexander and Michael Bronstein, and with the help
of lab engineer Eyal Gordon, we built a system that can distinguish between two identical twins
(Michael and Alexander) based on 3D pictures of their faces.  Our system can recognize faces under 
various poses and facial expressions.

In this talk I will review the components of our system and some theoretical problems and technical
challenges. The numerical building blocks include:
1. Kimmel-Sethian fast marching on triangulated domains (FMTD) scheme, that efficiently computes geodesic
distances on triangulated surfaces.

2. Elad-Kimmel bending invariant signatures for isometric surfaces.

3. Bronstein^2-Kimmel eigen-forms for expression invariant 3D face recognition.