Deep Learning-Powered Secure Multimodal Biometric Feature Fusion with Explainability and Real-Time Deployment

DOI

10.2991/978-94-6239-616-6_71How to use a DOI?

Keywords

Multimodal biometrics; reversible embedding; deep learning; Frequency-ReLU; cross-modal attention; explainable AI; SHAP; LIME

Abstract

Unimodal biometric systems have low discriminability, spoofing weaknesses, and noise sensitivity. This research proposes a secure, reversible, and comprehensible multimodal biometric verification architecture that integrates facial, fingerprint, and palmprint characteristics in order to address these issues. To incorporate all three biometrics into a single container image without compromising perceptual quality, an enhanced Least Significant Bit (LSB) reversible embedding approach is suggested. To improve latent representation learning and reconstruction quality, a convolutional autoencoder based on the Frequency-Domain Rectified Linear Unit (F-ReLU) is created. To dynamically describe interactions across various modalities, a multi-head cross-modal attention fusion network is shown. To provide trust and transparency, explainability is integrated utilizing SHAP and LIME to quantify local and global feature contributions. Experiments show better performance than traditional fusion and ReLU-based systems, with PSNR of 41.72 dB, SSIM of 0.981, and verification accuracy of 98.45%. The suggested approach is appropriate for airport immigration control, defense access, and medical identity authentication because to its strong discriminability, truthful explainability, and high reversibility.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - S. Selvarani
AU  - M. Mary Shanthi Rani
PY  - 2026
DA  - 2026/03/31
TI  - Deep Learning-Powered Secure Multimodal Biometric Feature Fusion with Explainability and Real-Time Deployment
BT  - Proceedings of the International Conference on Artificial Intelligence and Secure Data Analytics (ICAISDA 2025)
PB  - Atlantis Press
SP  - 950
EP  - 977
SN  - 1951-6851
UR  - https://doi.org/10.2991/978-94-6239-616-6_71
DO  - 10.2991/978-94-6239-616-6_71
ID  - Selvarani2026
ER  -