Enhancing multimodal biometric security through fundamental transforms and modified sigmoid fusion

Enhancing multimodal biometric security through fundamental transforms and modified sigmoid fusion

연구 분야: Databases

학회: Cluster Computing

This paper presents an advanced multimodal biometric security scheme that leverages fundamental transforms and a nonlinear function to enhance data protection. The proposed system integrates face and iris biometrics. For face image encryption, it utilizes the Discrete Wavelet Transform (DWT), Fourier Transform (FT), and Singular Value Decomposition (SVD). Additionally, a confusion-diffusion architecture is employed for iris image encryption. The Modified Sigmoid Function (MSF) is used to combine encrypted biometric patterns, producing a secure and revocable template. This fusion process improves the separation of features between individuals, improving the distinction between genuine users and fraudsters, and the robustness of the system. The system's efficacy is evaluated using correlation analysis, Receiver Operating Characteristic (ROC) curves, and Equal Error Rate (EER), demonstrating superior performance compared to existing methods. Our approach achieves an EER of zero and an AUC of one, outperforming conventional techniques by up to 15% in accuracy. This work highlights the role of nonlinear functions in improving the security and robustness of biometric systems, paving the way for future advancements in this field.