Building a Fingerprint Recognition Structure as a Security Mechanism to Verify the Identities of Users in Cloud Environment through Resnet-18

Objectives: To authenticate users employing fingerprint recognition thereby strengthening a security mechanism in a cloud environment. Methods: This study presents a novel approach for fingerprint detection in cloud systems by combining deep learning with metaheuristic optimization. A novel approach based on ResNet-18 for feature extraction and Thermal Exchange Optimization (TEO) classification is presented. This method aims to reduce data duplicity and provide safe access rights. When tested on 80% of the NIST and BDAL datasets, the proposed ResNet-18-TEO method outperforms existing models concerning recognition rates of 95.3% and 94.3% respectively. Findings: The experimental results reveal a False Acceptance Rate (FAR) spanning 0.6% to 4.8% and a False Rejection Rate (FRR) spanning 1.1% to 5.3%. Novelty: The proposed method would help cloud-based security since its fingerprint recognition time is substantially less than 25 seconds in training and 15 seconds in testing that of existing processes. In future work, this research will improvise using several deep learning techniques which are adopted using real time or research dataset. Keywords: Fingerprint recognition, Cloud, Security, Authentication, Privacy