Impact of Existing Deep CNN and Image Descriptors Empowered SVM Models on Fingerprint Presentation Attacks Detection

Abstract

Automatic Fingerprint Recognition Systems (AFRS) are the most widely used systems for authentication. However, they are vulnerable to Presentation Attacks (PAs). These attacks can be placed by presenting an artificial artifact of a genuine user’s fingerprint to the sensor of AFRS. As a result, Presentation Attack Detection (PAD) is essential to assure the security of fingerprint-based authentication systems. The study presented in this paper assesses the capability of various existing Deep-Learning and Machine-Learning models. We have considered four state-of-the-art Convolutional Neural Network (CNN) architectures such as MobileNet, DenseNet, ResNet, VGG as well as Support Vector Machine (SVM), trained with image descriptor features in our study. The benchmark LivDet 2013, 2015, and 2017 databases are utilized for the validation of these models. The experimental findings indicate toward the supremacy of Deep CNN models in cross-material scenario of PAs. © 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.