Multi-modal Biometric System for Face and Fingerprint using Convolutional Neural Network

Security issues are becoming more common as digital products and technology expand, and authentication technology is becoming increasingly important. Authentication has typically been accomplished through the use of ID cards, passwords, and PINs. It has grown in popularity over the years as a result of its use in areas like airports, secure financial transactions, banking, and mobile and computer access. Biometrics is a way to measure a person’s physical characteristics to authenticate their identity. Though these uni-modal biometric systems are more secure than the previous traditional approaches. Unimodal biometric systems are unable to handle situations including noisy data and non-universalities, inter-class variations, intra-class variations, and spoof attacks. Some of the problems can be solved by implementing multi-modal biometrics. It has become necessary due to recent technological advances to apply multi-modal biometrics to everyday applications. We present a multi-modal technique in which fingerprint and facial characteristics are integrated at fully connected layers. Using an early fusion strategy we fine-tune the hyper parameters of a designed network. According to the results of the experiments, the multi-biometric system utilizing the proposed level of fusion outperforms uni-biometric systems or systems using the existing level of fusion methodologies.