Signal and Image Processing for Biometrics

Chapter 4Modeling, Reconstruction and Trackingfor Face Recognition ¹

 

 

 

With the development of biometric techniques, automatic identity control systems have been invested in many places and facilities (e.g. airports and secure premises) during the last years. To improve the traffic flow at these recognition systems, it is necessary to minimize as much as possible the constraints imposed on the user. To meet this objective, it is necessary to perform “on-the-fly” acquisitions, without constraining the user to stop and stand in front of a sensor. In this chapter, we therefore focus on the use of facial biometrics, and more specifically on issues related to on-the-fly face acquisition. To enable the authentication within such systems, we have to solve a number of issues related to the facial shape and texture estimation. To address the theoretical aspects related to face acquisition and reconstruction, we consider the following framework: a multiview acquisition system is placed at the entrance of a room, a corridor, or a car park, for instance, and the aim is to identify or authenticate the person observed through this device.

4.1. Background

The requirements of a biometric system are varied, for example its ease of use, its speed of execution, its non-intrusiveness to users, its cost, and its reliability. The first three points are essential for systems designed for the general public, where the number of users is important and where these users are not specifically trained ...