Attack Detection Using Deep Learning–Based Multimodal Biometric Authentication System
Nishant Kaushal, Sukhwinder Singh, Jagdish Kumar · 2021
Biometrics-based human identity confirmation and management systems have gained popularity in recent years. Biometric traits of an individual ranging from fingerprint, palm print, iris, and face to even gestures have found applications in various devices such as smartphones, and laptops, and even its use has been seen in extensive scale management system, for example, “Aadhaar” in India. As biometrics-based authentication systems have increased, the issue of identity theft has also grown drastically. Biometrics-based authentication systems involve human-computer interaction, as the trait captured via a sensor is processed and then the identity is verified. The most common method used by hackers for identity theft is by using spoofs (presentation attack). Unimodal biometrics-based systems are more vulnerable than multimodal as they involve only one trait than multimodal in which two or more than two traits are used for identity authentication. A multiple expert-based decision-level fused detector has been designed and analyzed for an efficient and secure multimodal biometric-based human identity authentication system.