Signal and Image Processing for Biometrics

Chapter 3Facial Soft Biometrics for Person Recognition ¹

This chapter presents the idea of using soft biometric systems for person verification and identification. Such systems employ multiple traits for human identification, and do so in a non-intrusive and computationally efficient manner, as well as allow for fast and enrollment-free biometric analysis, even in the absence of consent and cooperation of the surveillance subject.

Motivated by the emergent need for robust soft biometric systems, we present a basic reliability analysis of such systems and a statistical characterization of relevant system parameters, such as traits, trait-instances, and authentication group size. Toward understanding the capabilities and limitations of such systems, we consider different pertinent issues, such as misclassification due to the image processing algorithms and cross-subject interference corresponding to the issue where different subjects share similar facial and anatomical characteristics.

3.1. Introduction to soft biometrics

Traditional biometrics offer a natural and reliable solution for establishing the identity of an individual, and for this reason, the use of human physical and behavioral characteristics have been increasingly adopted in security applications. With this approach maintaining various advantages, such as universality, robustness, permanence, and accessibility, it is not surprising that current security mechanisms and systems, be it for government or commercial facilities, ...