Signal and Image Processing for Biometrics

Chapter 10Hidden Biometrics1

 

 

 

10.1. Introduction

As we can see in the previous chapters, the different biometric modalities that have been considered, so far, deal essentially with visible human characteristics. Here, visible human characteristics refer to any part of the human body that can be visually identified. For instance, face, iris, hand, fingerprints, ears, etc., are visible characteristics that can be subject to forgeries. In such a case, we refer to “spoofing”. In fact, biometric systems can be spoofed when an imposter attempts to use, for example, a fake face, fake hand, or fake fingerprint to be identified as another person. Within this context, several research teams are working on this problem, and several solutions based on image and signal processing have been proposed. For example, to detect and differentiate a real face from a simple photo presented to a capture system, the analysis of facial vibrations and eyes blinking can be considered. Furthermore, multispectral illumination techniques can also detect fakes by considering the fact that physical properties (e.g. light reflection) of materials are different compared to those of living organs. Generally speaking, anti-spoofing systems can be reliable, but bypassing them is not an impossible task.

In this chapter, we discuss a biometric concept that is particularly robust regarding forgeries. More specifically, instead of considering human visible characteristics, we rather consider intrinsic, non-visible ...