TRIOSECURE: An Encoded Image-Based Biometric Authentication Framework

Kannan M, Rinsa Rees

Abstract: TrioSecure is a proposed image-based biometric authentication framework designed to address the critical limitations of existing biometric systems, including privacy leakage, spoofing attacks, and lack of explainability. Unlike conventional systems that store raw biometric images, TrioSecure converts biometric data from face, fingerprint, and iris modalities into encoded pixel representations using a custom encryption algorithm, thereby eliminating direct exposure of sensitive biometric data. Authentication is governed by a trio-verification mechanism comprising encoded feature comparison, Large Language Model (LLM)-based adversarial anomaly detection, and Natural Language Processing (NLP)-based semantic validation. LLMs analyze multimodal embeddings and associated metadata to generate human-readable explanations for each authentication decision, providing transparency and auditability. The system supports real-time encoding, spoof detection against advanced threats such as deepfakes and replay attacks, and explainable authentication outcomes. Built using Python, OpenCV, NumPy, and Flask, TrioSecure offers a secure, scalable, and cost-effective solution for modern identity verification systems that demand both high security and user-interpretable decision-making.

Keywords: TrioSecure, Biometric Authentication, Pixel Encoding, LLM-Based Spoof Detection, Explainable AI, Multimodal Biometrics, Privacy-Preserving Authentication, Adversarial Attack Detection, NLP Explanation Generation, Template Protection

How to Cite?: Kannan M, Rinsa Rees, "TRIOSECURE: An Encoded Image-Based Biometric Authentication Framework", Volume 15 Issue 4, April 2026, International Journal of Science and Research (IJSR), Pages: 1767-1772, https://www.ijsr.net/getabstract.php?paperid=SR26428102234, DOI: https://dx.dx.doi.org/10.21275/SR26428102234

Download Citation: APA | MLA | BibTeX | EndNote | RefMan