Description
Date depot: 1 janvier 1900
Titre: FACIAL SOFT BIOMETRICS. METHODS, APPLICATIONS AND SOLUTIONS.
Directeur de thèse:
Jean-Luc DUGELAY (Eurecom)
Domaine scientifique: Sciences et technologies de l'information et de la communication
Thématique CNRS : Non defini
Resumé:
Soft biometrics impart systems substantial advantages, to name a few: they do not need enrollment, their acquisition is non obtrusive and does not require the cooperation and consent of an observed individual. Those benefits come along with limitations, viz. the lack of distinctiveness and permanence.
To overcome these limits we construct a soft biometric system combining detection and classification of various soft biometric traits: hair, skin and eye color, beard, mustache and glasses.
First of all we study algorithms for the extraction of these traits in digital images. Furthermore we apply our system in the context of different applications. We emphasize on the scenario of person recognition using a soft biometrics system. We elaborate associated pertinent parameters and aspects, as of the number of traits, traits instances, size of authentication group, number of over all categories, etc.
We furthermore explore the use of such a soft biometric system for facial aesthetics evaluation.
Doctorant.e: Dantcheva Antitza