Face in Person Re-Identification

Face in Person Re-Identification

Andrea F. Abate (University of Salerno, Italy), Stefano Ricciardi (University of Salerno, Italy) and Genoveffa Tortora (University of Salerno, Italy)
Copyright: © 2014 |Pages: 16
DOI: 10.4018/978-1-4666-5966-7.ch013


The face represents one of the most diffused and established biometrics for both identity verification and recognition with a large corpus of research focused on advancing the accuracy, the robustness, and the response speed of face recognition systems by means of 2D, 3D, and hybrid approaches. One of the new research lines emerging in this field during the last years is face-based people re-identification, namely the task of recognizing new occurrences of an individual's face once it has been detected and initialized at a given time on the same location or eventually at other locations covered by a network of non-overlapping cameras. In this chapter, the main issues and challenges specifically related to face-based people re-identification are described, and the most promising techniques and results proposed on this topic so far are presented and discussed.
Chapter Preview

From Face Recognition To Face Re-Identification: Issues And Challenges

As already recalled before, the research on face recognition conducted in the last two decades produced a great number of algorithms and methodologies (Belhumeur et al., 1997; Etemad & Chellappa, 1997; Wiskott et al., 1997; Moghaddam, 2000; Bartlett et al., 2002, Wright et al., 2009). The first objective was mainly in raising the upper limit of recognition accuracy in controlled conditions (one of the explicit goals of FRGC) also because most of the first publicly available reference datasets for face recognition like the FERET (Phillips et al., 1998) and the YaleB (Georghiades et al., 2001) were acquired in studio with controlled settings and cooperative subjects.

Complete Chapter List

Search this Book: