Enhancing Multimodal Biometric Authentication Using Bi-LSTM Deep Learning Network
Laxman Singh, Ashish Kumar, Richa Golash · 2025
In recent years, biometric systems have gained immense popularity for secure authentication and identification due to their ability to capture unique physiological and behavioral traits. However, unimodal biometric systems often face challenges such as noise, intra-class variations, and spoofing attacks. To address these limitations, multi-modal biometric systems have emerged as a robust solution by combining multiple biometric traits, such as fingerprints, face, iris, and voice, which enhances accuracy and reliability. This paper presents a novel multi-model biometric system leveraging deep learning (DL) techniques. To classify data from various biometric sources, the suggested framework uses Bi-directional Long Short-Term Memory (Bi-LSTM). When developing the multi-model biometrics system (Face-Iris-Palm print), they consulted the MULB database. The experimental outcome was determined by several performance matrices, including,, , and. The results establish that the recommended model achieved the best results in terms of multi-model biometric system performance, with an of 98.91%, of 98.11%, of 97.07%, and of 98.89%.