The human face is the most well-researched object in computer vision, mainly because (1) it is a highly deformable object whose appearance changes dramatically under different poses, expressions, and, illuminations, etc., (2) the applications of face recognition are numerous and span several fields, (3) it is widely known that humans possess the ability to perform, extremely efficiently and accurately, facial analysis, especially identity recognition. Although a lot of research has been conducted in the past years, the problem of face recognition using images captured in uncontrolled environments including several illumination and/or pose variations still remains open. This is also attributed to the existence of outliers (such as partial occlusion, cosmetics, eyeglasses, etc.) or changes due to age. In this chapter, the authors provide an overview of the existing fully automatic face recognition technologies for uncontrolled scenarios. They present the existing databases and summarize the challenges that arise in such scenarios and conclude by presenting the opportunities that exist in the field.
TopIntroduction
Unveiling the way humans perceive identities has been of great interest to psychologists for at least five decades (Bruner & Tagiuri, 1954). It nowadays constitutes one of the most popular research areas in experimental psychology (Bruce & Young, 1986; Sinha, Balas, Ostrovsky, & Russell, 2006). The interested reader can find a nice summary of research findings regarding human perception of identities in (Sinha, Balas, Ostrovsky, & Russell, 2006).
Automatic face recognition was first attempted in the 1960s (Bledsoe, 1964) and 1970s (Kelly, 1970). A pioneer in the field was Takeo Kanade (1973), who first attempted facial features localization in order to match them for face recognition. Since then, face recognition became one of the mainstream applications of image analysis and computer vision, creating a wealth of scientific research (the interested reader may refer to Chellappa, Charles, and Saad (1995) and Zhao, Chellappa, Phillips, and Rosenfeld (2003) to see the progress of face recognition during the past 30 years and until 2003).
The collection of face databases, among the first large object databases that were created, made face recognition the main application for certain domains of statistical machine learning and pattern recognition and particular for the domain of component analysis (CA). CA includes methods such as Principal Component Analysis (Turk & Pentland, 1991; Kirby & Sirovich, 1990), Linear Discriminant Analysis (Belhumeur, Hespanha, & Kriegman, 1997), Independent Component Analysis (Bartlett, Movellan, & Sejnowski, 2002) and Nonnegative Matrix Factorization (Zafeiriou, Tefas, Buciu, & Pitas, 2006).
Face recognition can be further distinguished into a set of sub-problems, each of which dealing with the different challenges met by using different problem formulations according to the machine learning and matching algorithms it employs: