Sequential Multimodal Biometric Authentication using Eye, Voice, and Gesture Verification
- Version
- Download 24
- File Size 450.89 KB
- File Count 1
- Create Date 19 February 2026
- Last Updated 19 February 2026
Sequential Multimodal Biometric Authentication using Eye, Voice, And Gesture Verification
Shivani, Mitali Khujnare, Diksha Pawar, Shraddha Pore
Sandip University, SOCSE, B.Tech CSE (AIML) [Nashik, India]
Abstract
Modern biometric authentication systems operating on single modalities remain vulnerable to presentation and replay attacks. While multimodal biometrics improves reliability, most existing systems employ score-level fusion where strong modalities compensate for weak
ones, preserving attack pathways. This paper proposes a sequential decision-level multimodal authentication framework integrating eye behavior, voice dynamics,and gesture motion patterns using commodity sensors.The system enforces non-compensatory verification,requiring all modalities to independently validate identity. Experimental evaluation involving controlled impostor attempts shows individual false acceptance rates of 7.8%, 6.9%, and 9.2% for eye, voice, and gesture modalities respectively, while the combined system reduces false acceptance to 0.32%. Theoretical analysis predicts an attack probability of 0.049%, closely matching empirical observations. The results demonstrate that sequential multimodal authentication significantly improves security without specialized hardware, making it suitable for practical secure access systems.
Index Terms— Multimodal biometrics, authentication security, liveness detection, behavioral biometrics, gesture recognition, speaker verification.
Download