Graphics Reference
In-Depth Information
5
Applications
Stefano Berretti
1
, Boulbaba Ben Amor
2
, Hassen Drira
2
, Mohamed
Daoudi
2
, Anuj Srivastava
3
, Alberto del Bimbo
1
, and Pietro Pala
1
1
Dipartimento di Sistemi e Informatica, University of Firenze, Firenze, Italy
2
Institut Mines-Telecom/Telecom Lille 1, France
3
Department of Statistics, Florida State University, Tallahassee, USA
5.1 Introduction
In the previous chapters, techniques that can effectively represent the surface of 3D face scans
to perform face analysis were presented. These techniques were mainly developed to perform
face identification and verification to enable provison of complementary and/or alternative
solutions with respect to traditional 2D approaches. In fact, a lot of research effort has been
invested on this topic in the last few years, resulting in the development of many different
techniques that help obtain very high recognition accuracy on benchmark data sets. However,
as 3D face recognition methods become more robust, effective, and efficient, new tasks are
emerging where 3D techniques can be usefully applied.
The main objective of this last chapter of the topic is to give an overview of emerging
3D face processing applications and new trends of research interest. In particular, in the next
sections of the chapter, first an overview of the 3D face databases that are released for public
use are presented, evidencing their main characteristics in terms of expression variations, pose
changes, presence of occlusions (Section 5.2), then different applicative scenarios will be
presented, as follows:
In Section 5.3, state-of-the-art methods for 3D facial recognition are reviewed and discussed.
•
In Section 5.3.3, the challenges related to face recognition using 3D scans with non-frontal
pose, missing parts, and occlusions are discussed, and methods that address this applicative
scenario are reviewed. More insights are given for two of these methods.
•
