An Efficient Multimodal Biometric Key Binding Technique for Secure Authentication

Abstract

Cryptography plays a vital role in protecting the data using cryptographic key. Due to rapid growth of technology in recent years, biometric cryptosystems have become more prevalent. But protecting the biometric key is a big challenge which can be achieved by several key binding techniques. Biometric key binding technique can be implemented using unimodal or multimodal biometric traits. Research proves that multimodal biometric have high performance than unimodal biometric system. In multimodal biometric system, fusion techniques are used to combine the biometric features of different traits. Mathematical methods are used for fusion of traits which are irreversible.  Thus, multimodal biometric such as face and fingerprints are used for implementation. In our paper, feature extraction is performed by transfer learning and traits are fused by additive function. The secret key of size 128 bits is fused by an effective key binding technique with random permutation finally accuracy is compared with the existing methods.