Personalized CNN Architecture for Advanced Multi-Modal Biometric Authentication
Rashmi A. Joshi, Nitin B. Sambre · 2024
This study presents a novel biometric recognition model amalgamating fingerprint and iris imaging modalities for enhanced identity verification. Through a comparative analysis with standard CNN architectures, including VGG16, VGG19, ResNet-50, ResNet-101, DenseNet-121, Inception-V3, and MobileNet-V2, the proposed model emerges as a robust solution. Achieving peak accuracy of 98.5% with 128 neurons in the second last dense layer and 600 epochs, the model demonstrates superior performance. The feature extraction using customized Convolutional Neural network (CNN) is done. The output of CNN is given to and Long Short-Term Memory network (LSTM) based model. This model contributes to its adaptability and effectiveness across diverse datasets. This innovative approach positions the model as a promising advancement in biometric recognition, offering heightened security and accuracy in identity verification processes.