SURFBCS: speeded up robust features based fuzzy vault scheme in biometric cryptosystem

Biometric cryptosystem (BCS) is an emerging research area in which Biometric alleviates the need to remember passwords and cryptography provides a sense of security. The use of biometrics in cryptography alleviates the need to remember passwords. This paper proposes a novel method called Speeded Up Robust Features (SURF) based Biometric Cryptosystem (SURFBCS) for biometric key generation. This method makes use of a well-known Fuzzy vault scheme for data security. The motivation behind using SURF scheme for biometric cryptosystem is that the obtained features are invariant to rotation and scaling which is essential for improved authentication. SURFBCS is divided into three stages, i.e., Key generation, Enrollment and Authentication. In key generation stage, biometric features are extracted using Speeded Up Robust Features which are subsequently used to generate the biometric key. The enrollment stage uses these features and biometric key to create a fuzzy vault. The authentication stage uses this fuzzy vault to authenticate the query biometric. Extensive experiments have been conducted on five ear databases, viz., AMI, CP, IITD-v1, IITD-v2 and USTB-v2. The performance of the proposed method is evaluated in terms of accuracy, false reject rate (FRR), vault construction time, key recovery time and Receiver Operating Characteristics (ROC) curve and compared with the state-of-art methods. The highest accuracy of 94.9% is obtained by SURFBCS method on CP database. Moreover, vault construction and key recovery time of 31.69 ms and 0.90 ms, respectively, is obtained by the proposed method on IITD-v1 database also supersedes the other methods. It has been observed that the proposed method is better than several state-of-the-art methods in both qualitative and quantitative terms.