Local feature encoding for unknown presentation attack detection: An analysis of different local feature descriptors

Abstract

In spite of the advantages of using fingerprints for subject authentication, several works have shown that fingerprint recognition systems can be easily circumvented by means of artificial fingerprints or presentation attack instruments (PAIs). In order to address that threat, the existing presentation attack detection (PAD) methods have reported a high detection performance when materials used for the fabrication of PAIs and capture devices are known. However, for more complex and realistic scenarios where one of those factors remains unknown, these PAD methods are unable to correctly separate a PAI from a real fingerprint (i.e. bona fide presentation). In this article, a new PAD approach based on the Fisher Vector technique, which combines local and global information of several local feature descriptors in order to improve the PAD generalisation capabilities, was proposed. The experimental results over unknown scenarios taken from LivDet 2011 to LivDet 2017 show that our proposal reduces the top state-of-the-art average classification error rates by up to four times, thereby making it suitable in real applications demanding high security. In addition, the best single configuration achieved the best results in the LivDet 2019 competition, with an overall accuracy of 96.17%.This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

A comprehensive survey of fingerprint presentation attack detection

“…Ghiani et al [78] conducted experiments with features based on pore detection, ridge wavelets and several textural features. They concluded that the best method was LBP, although it has the disadvantages of being sensitive to image rotation and the need for longer computational time due to its long histogram.…”

“…Researchers have found that the textural features [Figure 7] of the fingertip, such as smoothness and morphology, can be used to distinguish real fingerprints from artificial ones [78,79] . Methods belonging to this category make use of such textural features.…”

On the Effectiveness of Impedance-Based Fingerprint Presentation Attack Detection

“…For this purpose, the SVM [ 52 ] is used as a classifier, since it constantly achieved remarkable performances across different fingerprint PAD studies [ 34 , 35 , 53 , 54 , 55 , 56 , 57 ]. The SVM is designed to work on high-dimensional input data and derive binary decisions by defining a hyperplane that separates both classes.…”

“…However, when focussing on unknown attacks and cross-sensor and cross-database scenarios, it is clear that handcrafted methods are able to outperform deep learning approaches, as was shown by the winner of the LivDet 2019 competition [34]. This work was further extended in [35] to include even more features into the fisher vector encoding before classifying these with a support vector machine (SVM). In the area of deep learning, Chugh and Jain [40] proposed their Fingerprint Spoof Buster as a patch-based convolutional neural network (CNN) together with two datasets, MSU-FPAD and PBSKD.…”

On the generalisation capabilities of Fisher vector‐based face presentation attack detection

“…To tackle those issues, we focus on a different approach that has shown remarkable results for challenging scenarios such as cross‐dataset and unknown PAI species for fingerprint PAD [21, 22]. That work used a combination of local feature descriptors and global feature representation models with a new feature space in which the generalisation capabilities of the PAD module are enhanced.…”