Face Recognition

A biometric system based on face recognition has a very large number of application: security systems, criminal identification, image and movie processing, man-machine interaction, ... Unfortunately the development of computational models for the face recognition is a very difficult task; today we do not even know how the human brain recognize the faces. The automatic face recogniton involves the resolution of some complex problems :

  • face localization in complex scenes
  • invariance to pose and illumination
  • invariance to change in expression
  • invariance to moustache, beard, glasses, hair style, ...

Three main stages constitute the core of our approach: Approximate location, Fine location and Face verification

The first stage approximately localizes all the elliptical objects within a certain range of variation present in a directional image. This is effected through the implementation of a Generalized Hough transform, which uses sectors of an elliptical annulus as templates.

The second stage improves the localization accuracy through a local optimization of the ellipse position and size. Finally, the third stage checks whether the objects found are faces or not. To this purpose, the projection method can be applied directly to the directional images instead of to the original ones: in fact it is easy to prove that local maxima and minima are present in the eye, nose and mouth regions due to the presence of horizontal and vertical directions.