MULTI-MODAL BIOMETRIC AUTHENTICATION SYSTEM USING DEEP NEURAL NETWORKS

Abstract

The issue of giving authorized owners of a certain identification process safe and easy access to information and solutions is known as identity management. The primary goal in ascertaining an individual's identity is the implementation of the secured identification feature. Tokens, access cards, PINs, keys, and passwords are a few examples of private identifying components that are used often yet can be lost, stolen, cracked, copied, or posted. It is essential to comprehend biometrics-based identification in order to prevent the drawbacks. Not only can intracategorical changes affect nonuniversality, but they also affect sound and false strikes. Multimodal biometrics are utilized to remove the episodes, which are essentially a collection of countless modalities. Fingerprints and palmprints can be used as sources of authentication. Using this technique, rich neural communities (DNN) were surely projected for fine-grained attribute extraction and item detection. Multimodal biometric solutions are designed to fulfill the rigorous needs of the industry due to the limits of unimodal biometric structures, which lead to high delivery demands, restricted talent splitting, and considerable False Acceptance Rates (FAR) and False Rejection Rates (FRR). In reality, values of Euclidean distance are used for tiny details. The suggested method is highly safe and attains a better recognition speed just while dealing with loud issues.