Konferenzbeitrag
Remote Cancelable Biometric System for Verification and Identification Applications
Lade...
Volltext URI
Dokumententyp
Text/Conference Paper
Dateien
Zusatzinformation
Datum
2023
Autor:innen
Zeitschriftentitel
ISSN der Zeitschrift
Bandtitel
Quelle
Verlag
Gesellschaft für Informatik e.V.
Zusammenfassung
Cancelable biometric schemes protect the privacy of biometric templates by transforming them, with the help of a key, into an irreversible form that can be replaced if compromised. While these schemes provide more advantages in the user-specific key setting, their application with the user-specific key setting is limited in the identification scenario. Alternatively, the application-specific key setting can be used to employ cancelable biometric systems for the identification scenario. However, in an application-specific key setting, cancelable biometric schemes become static with respect to the protected template replacement; if a protected template or the key is compromised, then the replacement of all the protected templates stored within the same application is mandatory. In addition, experimental results show a degradation of performance for the application-specific key setting in cancelable biometric systems. In this paper, we consider a remote recognition protocol based on cancelable biometric schemes in the identification and verification scenarios so that trusted users can generate protected templates and send them to a server. The server can compare the protected query with the protected templates enrolled in the database for recognition. We investigate the user-specific key setting for cancelable biometric schemes for both verification and identification scenarios, which provides those systems with a dynamic replacement of compromised templates. In our experiments, we analyze different cancelable biometric schemes, including BioHashing, Multi-Layer Perceptron (MLP) Hashing, and Index-of-Maximum (IoM) Hashing. We evaluate their performances when applied within our proposed protocol for face recognition and speaker recognition on the IARPA Janus Benchmark C (IJB-C) and NIST-SRE04-16 datasets for user-specific key and application-specific key settings. The source code of all our experiments is publicly available to facilitate the reproducibility of our work.