FT-HT: A Fine-Tuned VGG16-Based and Hashing Framework for Secure Multimodal Biometric System

Abstract

Multimodal biometric systems offer several advantages over unimodal systems, including a lower error rate, greater accuracy and broader coverage of residents. However, the multimodal systems need to store multiple biometric traits associated with each user, which brings a higher need for integrity and privacy. This study describes a deep learning (DL) model for a feature-level coalition that utilizes the biographical data of the user's face and iris to create a secure multimodal template. To create a reliable, unique multimodal shareable latent image, a deep hashing (linearization) approach is used for the fusion architecture. Furthermore, a hybrid secure architecture that fuses secure sketching techniques with erasable biometric features and integrates them into a complete security framework is used in this work. The efficiency of the recommended method is demonstrated using the face and iris images from the multimodal database. The proposed method provides the ability to delete templates and better protect the biometric data. This method works with the “WVU” multimodal data store and the “hashing” method for “image retrieval.” The proposed improved VGG16 achieves a data accuracy of 99.85. The paper also provides information on the techniques for structuring modalities such as iris and face using deep hashing, multimodal fusion and biometric security techniques. However, further studies are needed to extend the proposed framework to other unrestricted biometric aspects.