Face Analysis Under Uncontrolled Conditions

4Extraction of Facial Features

Benjamin ALLAERT¹, Ioan Marius BILASCO² and Chaabane DJERABA²

¹IMT Nord Europe, Lille, France

²University of Lille, France

4.1. Introduction

Facial analysis is a growing field of research as it concerns many applications such as security (biometrics, surveillance), robotics (human–machine interaction) or telecommunications (affective analysis). These systems have the potential to be easily deployed and are considered not to be very invasive compared to other biometric systems (fingerprints, iris recognition, etc.), and their use is gaining popularity with the evolution of communication tools (Internet, mobile).

Systems for face characterization generally use very similar processes. Figure 4.1 represents the different steps of a generic facial analysis process. The first step is to detect the face in an image. Once the face is detected, information such as texture and facial geometry are extracted. A normalization step is sometimes necessary to reduce the difference between intra- and inter-individual faces, in particular in the presence of head pose variations, lighting changes and occlusions (e.g. glasses). Once the extraction of facial information is completed, the faces are labeled by a classification system according to the expected analysis.

In the following, we discuss in detail the different steps of the facial analysis process. For each of these steps, we present the most representative methods in the literature. We conclude this chapter ...