Federal University of Pernambuco
This paper introduces a new multi-instance key regeneration system used to regenerate cryptographic keys from biometric data. The serial concatenation codes is used with a single mechanism that improves the biometric performance and security of the system, also making possible the regeneration of longer and higher entropy cryptographic keys. The system was evaluated on two public databases: CBS and NIST-ICE 2005. On NIST-ICE 2005 it is possible to regenerate a 287 binary digit cryptographic key with estimated entropy of 160 bits at 0% False Acceptance Rate (FAR) and 0.3371% False Rejection Rate (FRR).