ABSTRACT
Biometric template protection has gained a lot of importance in recent years due to advances in technology and rise of Internet of Things. Almost everything ranging from house doors to bank lockers, mobile phones to nuclear reactors, class attendance to passport, happens to be controlled by biometrics for secure access. Biometric authentication has touched almost every person’s life in some form or the other, yet we do not come across systems that ensure the safety and privacy of collected templates. Various biometric template protection techniques have been proposed in this regard, and the most popular ones are Fuzzy Vaults, Fuzzy Commitment, Biometric Cryptosystems, and Cancelable Biometrics. Amongst all the existing schemes, cancelable biometrics enriches the protection environment with important qualities of revocability and diversity, while enhancing users’ control over their right to privacy with respect to biometric templates. The fascinating concept of cancelable biometric which transforms a biometric to its pseudo-version was proposed in 2001. Twenty years since its inception, the concept has been studied, developed, and analysed in numerous ways. Still, it has yet to achieve the realms of real-life implementations. This chapter aims to tap the ground problems and their possible solutions like compromise of entire database, problem of multiple identity registrations, and realising issuance and distribution of user-specific tokens, so that the concept may be utilised to its full potential. It also reviews the recent contribution and advances of neural networks concepts like convolutional neural networks and autoencoders in the template protection domain and scope of these techniques in cancelable biometrics.
