Anti-spoofing Approach Using Deep Convolutional Neural Network

Abstract

Our research aims at classifying biometric image samples from well-known spoofing databases that encompass images with different resolution and sizes using a deep Convolutional Neural Network (CNN). In this effort, we optimally use the CNN for biometric image classification to prevent spoofing attack in an extensive range. This work detects the presentation attacks on facial and iris images using our deep CNN, inspired by VGGNet and Alex-Net. We applied the deep neural net techniques on three different biometric image datasets, namely ATVS, CASIA two class, and CASIA cropped. The datasets, used in this research, contain images that are captured both in controlled and uncontrolled environment along with different resolutions and sizes. We obtained the best test accuracy of 97% on ATVS Iris datasets. For CASIA two class and CASIA cropped datasets, we achieved the test accuracies of 96% and 95%, respectively.