Efficient Implementation of Cancelable Optical Face Recognition Based on Elliptic Curve Encryption

Abstract Most modern authentication systems adopt human biometrics to avoid any shortcomings that result from forgetting passwords and security codes utilized in traditional systems. To improve the security level of the original biometric traits from offensive attacks, cancelable biometric patterns are generated from the original biometric templates. This paper presents a new approach for cancelable face recognition based on the concept of elliptic curve cryptography (ECC). The ECC is classified as a public-key (asymmetric) encryption technique. In public-key encryption, each user (transmitter and receiver) has two keys: public key, and private-key. The proposed framework guarantees full distortion and encryption of the original biometric traits to be saved in the database to completely save the original biometrics away from intruders. To validate the performance of the proposed approacg, three face biometrics databases have been used. Receiver Operating Characteristic (ROC) curve and correlation scores are estimated to test the system quality. The simulation results prove that the proposed approach is efficient, robust, and it achieves high performance with three face databases.