Physiological signals in biometrics: an investigation on remote-photoplethysmography-based attack

In recent years, the unobservability of physiological signals has significantly improved the effectiveness of biometric authentication systems. In particular, photoplethysmography (PPG) signals are advantageous due to the ease of measurement and the low cost. However, the advent of remote PPG (rPPG) has enabled potential adversaries to acquire PPG signals through simple RGB videos. This thesis will compare the results obtained by using video PPG and finger PPG for biometric authentication. The aim of this work is to analyze the biometric-based authentication, with a particular focus on the rPPG method. First of all there will be an introduction to the definition of PPG and rPPG signals, followed by the extraction method used in the authentication. Then there will be an in-depth study of possible attacks to the authentication system, with particular attention to the Presentation Attack. In conclusion there will be a simulation of this attack using as measurement sites the fingertip and the rPPG signal extracted from video.