The natural instinct of every human being is to want to protect themselves from their surroundings. Starting from the classic passwords required to access specific information, we moved on to safer and more accurate methods based on the specific characteristics of each user: biometrics. The first identification using fingerprints, in fact, dates back to the last years of the 1800s and was used by the police station for the criminal identification. From that moment on, the various biometric authentication systems based on fingerprints began to spread and, consequently, the first bad guys began to create false fingerprints to access the systems themselves. Fingerprint spoofing techniques became effective against biometric sensors thus leading to the creation of Liveness Detector modules, capable of detecting the liveness of a fingerprint: from handcrafted methods to deep neural networks, the performances, tested on datasets of first editions of the LivDet competitions containing live prints and fake prints created with the consensual method, show high security. And no matter how much a "prey" may commit to defending itself, there will always be a "predator" ready to improve itself to reach its end. In this PhD thesis a new fingerprint falsification technique will be presented, able to show the vulnerabilities of the detectors presented in the last two editions of the International Fingerprint Liveness Detection Competition. This method based on latent fingerprints will be analyzed primarily from a pseudo-consensual point of view, to then move on to a completely non-consensual case study, simulating a real attack on a specific user. An adversarial perturbation technique via GAN will then be presented, in order to create, first digitally and then physically, a print that alters the result of the classification from fake to live. For this type of process the cross-sensor explainability will be studied, evaluating the performances step by step with the best detectors of the latest LivDet competitions.

The Art of Fingerprint Spoofing

CASULA, ROBERTO
2023-07-17

Abstract

The natural instinct of every human being is to want to protect themselves from their surroundings. Starting from the classic passwords required to access specific information, we moved on to safer and more accurate methods based on the specific characteristics of each user: biometrics. The first identification using fingerprints, in fact, dates back to the last years of the 1800s and was used by the police station for the criminal identification. From that moment on, the various biometric authentication systems based on fingerprints began to spread and, consequently, the first bad guys began to create false fingerprints to access the systems themselves. Fingerprint spoofing techniques became effective against biometric sensors thus leading to the creation of Liveness Detector modules, capable of detecting the liveness of a fingerprint: from handcrafted methods to deep neural networks, the performances, tested on datasets of first editions of the LivDet competitions containing live prints and fake prints created with the consensual method, show high security. And no matter how much a "prey" may commit to defending itself, there will always be a "predator" ready to improve itself to reach its end. In this PhD thesis a new fingerprint falsification technique will be presented, able to show the vulnerabilities of the detectors presented in the last two editions of the International Fingerprint Liveness Detection Competition. This method based on latent fingerprints will be analyzed primarily from a pseudo-consensual point of view, to then move on to a completely non-consensual case study, simulating a real attack on a specific user. An adversarial perturbation technique via GAN will then be presented, in order to create, first digitally and then physically, a print that alters the result of the classification from fake to live. For this type of process the cross-sensor explainability will be studied, evaluating the performances step by step with the best detectors of the latest LivDet competitions.
17-lug-2023
File in questo prodotto:
File Dimensione Formato  
TesidiDottorato_RobertoCasula.pdf

Solo gestori archivio

Descrizione: The Art of Fingeprint Spoofing
Tipologia: Tesi di dottorato
Dimensione 7.7 MB
Formato Adobe PDF
7.7 MB Adobe PDF   Visualizza/Apri   Richiedi una copia

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11584/371645
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact