Presentation attack detection in fingerprint biometrics

Abstract

There has been a rapid growth in services like finance which utilize fingerprint biometrics for user authentication. This has led to an increase in the need for a secure and reliable fingerprint recognition system to provide privacy and prevent fraud. We are proposing a novel end-to-end fingerprint presentation attack detection method based on the combination of Machine learning and Deep learning. In our proposed model, a Deep CNN architecture i.e. MobileNet v1 is used for the extraction of important features from input images while the actual classification takes place using a support vector machine. The proposed model is tested on LivDet 2013, 2015 and 2017 datasets and compared with vari ous state-of-the-art methods. Our model achieves an average accuracy of 98.88% in LivDet 2013, 96.74% in LivDet 2015 and 94.90% in LivDet 2017 datasets.