Join differential invariants for 3D face recognition

Relatore:  Marinella Cadoni - Università di Sassari
  martedì 21 aprile 2009 alle ore 16.00 16.00 inizio seminario, caffe' e pasticcini al termine

Invariants are widely used in Computer Vision to characterise clouds of points independently from the acquisition viewpoint. A well established mathematical theory, the so called “Moving Frame Theory” provides an algorithmic way to compute optimal sets of invariants that characterise a surface up to a Lie Group transformation.   

Following the outline of this theory, an optimal set of invariants to Euclidean motions can be generated and extended to the discrete case of point clouds, thus providing a methodology for their characterisation up to rigid motions and finally registration.
In this talk, an implementation of the methodology in the biometrical context of 3D face recognition and authentication showing promising results is presented.

Manuele Bicego

Data pubblicazione
17 aprile 2009

Offerta formativa