Multimodal Biometric System Fusion Using Fingerprint and Iris with Convolutional Neural Network

Abstract

Biometric sensing technology became everyday life frequent component as a result of world requirement for info security and safety legislation. A strong and efficient individual authentication has appeared because of new developments in multimodal biometrics. Multimodal biometrics integrates different biological traits in trying for creating considerable effect on identification performance. Latent fingerprint biometrics refer to effective human identification system for criminals given the accessible crime evidence shreds. Although, biometric trait restrictions like intra-class variation, sensed data noise, lack of individuality caused low matching score that possesses a negative effect on recognition and investigation process. This paper uses two unimodal biometrics—the fingerprint and the iris—applied as multi-biometrics to show that using these biometrics can produce excellent results with high accuracy. Every biometric result is weighted for involvement in the final decision, and the decision level is utilized for fusion. For every biometric result integration effect, a neural network is used. The datasets' experimental findings have demonstrated a notable biometric system identification capability. The accuracy performance of the suggested approach is 100, the FAR is 0.1, and the EER is 0.1. To demonstrate the efficacy of the suggested system, the suggested method is contrasted with a few other approaches currently in use.