Abstract
Multi-modal biometric systems (MBS) are an enhancement of unibiometric systems where recognition is based on multiple biological traits of the user. Once the biological traits of a user are captured through sensors, the unique feature points are extracted from each modality and represented as a feature vector. The privacy and security of feature templates in biometric recognition systems are the top design issues, which is gaining the attention of the majority of research community. In this paper, a hybrid template security technique is designed for a bi-modal biometric system based on fingerprint and hand geometry. The technique first use the bio-hashing, in order to transform the actual feature vectors (FVs) into respective binary feature vectors (BFVs) and then applying a transformation scheme to fuse the binary vectors into a secured template. In order to improve the storage overhead for the protected template, a novel octet indexing technique is applied on the intermediate feature vector (IFV). The proposed technique results in improvement of storage requirement by representing secured template as a 64-bits vector with octet indexing. Furthermore, the proposed technique results in the overall enhancement of the recognition accuracy as compared to fingerprint and hand geometry uni-biometrics. The proposed hybrid scheme results in a recognition rate of 98.4% and an equal error rate (EER) of 0.48%, with improvement in template storage overhead up to 50%.
Get full access to this article
View all access options for this article.
