Emanuela Marasco

A Survey on Antispoofing Schemes for Fingerprint Recognition Systems

Several issues related to the vulnerability of fingerprint recognition systems to attacks have been highlighted in the biometrics literature. One such vulnerability involves the use of artificial fingers, where materials such as Play-Doh, silicone, and gelatin are inscribed with fingerprint ridges. Researchers have demonstrated that some commercial fingerprint recognition systems can be deceived when these artificial fingers are placed on the sensor; that is, the system successfully processes the ensuing fingerprint images, thereby allowing an adversary to spoof the fingerprints of another individual. However, at the same time, several countermeasures that discriminate between live fingerprints and spoof artifacts have been proposed. While some of these antispoofing schemes are hardware based, several software-based approaches have been proposed as well. In this article, we review the literature and present the state of the art in fingerprint antispoofing.

Combining perspiration- and morphology-based static features for fingerprint liveness detection

It has been showed that, by employing fake fingers, the existing fingerprint recognition systems may be easily deceived. So, there is an urgent need for improving their security. Software-based liveness detection algorithms typically exploit morphological and perspiration-based characteristics separately to measure the vitality. Both such features provide discriminant information about live and fake fingers, then, it is reasonable to investigate also their joint contribution. In this paper, we combine a set of the most robust morphological and perspiration-based measures. The effectiveness of the proposed approach has been assessed through a comparison with several state-of-the-art techniques for liveness detection. Experiments have been carried out, for the first time, by adopting standard databases. They have been taken from the Liveness Detection Competition 2009 whose data have been acquired by using three different optical sensors. Further, we have analyzed how the performance of our algorithm changes when the material employed for the spoof attack is not available during the training of the system.

Fingerprint liveness detection based on histograms of invariant gradients

Security of fingerprint authentication systems remains threatened by the presentation of spoof artifacts. Most current mitigation approaches rely upon the fingerprint liveness detection as the main anti-spoofing mechanisms. However, liveness detection algorithms are not robust to sensor variations. In other words, typical liveness detection algorithms need to be retrained and adapted to each and every sensor used for fingerprint capture. In this paper, inspired by popular invariant feature descriptors such as histograms of oriented gradients (HOG) and the scale invariant feature transform (SIFT), we propose a new invariant descriptor of fingerprint ridge texture called histograms of invariant gradients (HIG). The proposed descriptor is designed to preserve robustness to variations in gradient positions. Spoofed fingerprints are detected using multiple histograms of invariant gradients computed from spatial neighborhoods within the fingerprint. Results show that proposed method achieves an average accuracy comparable to the best algorithms of the Fingerprint Liveness Detection Competition 2013, while being applicable with no change to multiple acquisition sensors.

Combining match scores with liveness values in a fingerprint verification system

We discuss the problem of combining biometric match scores with liveness measure values in the context of fingerprint verification. Recent literature has focused on the development of methods to assess if an input fingerprint sample is a “live” entity or a “spoof” artefact. This is commonly done by generating a single-valued numerical entity referred to as the liveness measure value. However, the problem of combining this liveness value with match scores has not been rigorously investigated. The goal of this work is to design a framework in which a liveness detector is incorporated with a fingerprint matcher. We first design and analyze three different methods to combine match scores with liveness values. Next, we introduce a Bayesian Belief Network (BBN) scheme that models the relationship between match scores and liveness values. Experiments carried out on a publicly available database of the Fingerprint Liveness Detection Competition 2009 (LivDet09) show the effectiveness of assuming a certain degree of influence of liveness values on match scores.

Increase the security of multibiometric systems by incorporating a spoofing detection algorithm in the fusion mechanism

The use of multimodal biometric systems has been encouraged by the threat of spoofing, where an impostor fakes a biometric trait. The reason lies on the assumption that, an impostor must fake all the fused modalities to be accepted. Recent studies showed that there is a vulnerability of the existing fusion schemes in presence of attacks where only a subset of the fused modalities is spoofed. In this paper, we demonstrated that, by incorporating a liveness detection algorithm in the fusion scheme, the multimodal system results robust in presence of spoof attacks involving only a subset of the fused modalities. The experiments were carried out by analyzing different fusion rules on the Biosecure multimodal database.

An anti-spoofing technique using multiple textural features in fingerprint scanners

Fingerprint verification systems may be circumvented by fake fingerprints produced using inexpensive materials like gelatin or silicon. An efficient countermeasure against these attacks is given by liveness detection. In the recent literature, different algorithms for detecting signs of vitality have been proposed. The cheapest techniques are software-based and utilize acquired fingerprint images in order to extract static or dynamic characteristics. In this paper, we propose a novel software-based solution for liveness detection based on static features coming out from the visual texture of the image. The reported results show that the use of our features effectively improves the discriminative power (between live and fake fingerprints) achieved by the algorithms proposed during the Liveness Detection Competition 2009.

Predicting identification errors in a multibiometric system based on ranks and scores

The goal of a biometric identification system is to determine the identity of the input biometric data. In such a system, the input probe (e.g., a face image) is compared against the labeled gallery data (e.g., face images in a watch-list) resulting in a set of ranked scores pertaining to the different identities in the gallery database. The identity corresponding to the best score is then associated with that of the probe. The aim of this work is to predict identification errors and improve the recognition accuracy of the biométrie system. The method utilizes the rank and score information generated by the identification operation in order to validate the output. Further, we demonstrate the proposed predictor can be effectively applied in multimodal scenarios. Experiments performed on two multimodal databases show the effectiveness of our framework in improving identification performance of biométrie systems.

Improving the Accuracy of a Score Fusion Approach Based on Likelihood Ratio in Multimodal Biometric Systems

Multimodal biometric systems integrate information from multiple sources to improve the performance of a typical unimodal biometric system. Among the possible information fusion approaches, those based on fusion of match scores are the most commonly used. Recently, a framework for the optimal combination of match scores that is based on the likelihood ratio (LR) test has been presented. It is based on the modeling of the distributions of genuine and impostor match scores as a finite Gaussian mixture models. In this paper, we propose two strategies for improving the performance of the LR test. The first one employs a voting strategy to circumvent the need of huge datasets for training, while the second one uses a sequential test to improve the classification accuracy on genuine users. Experiments on the NIST multimodal database confirmed that the proposed strategies can outperform the standard LR test, especially when there is the need of realizing a multibiometric system that must accept no impostors.

Robust and interoperable fingerprint spoof detection via convolutional neural networks

Fingerprint recognition for automated border control and other high-security applications needs robust integrated anti-spoofing capability. Facing the threat of presentation attacks, two key challenges to be solved are sensor interoperability and robustness versus new fabrication materials. This paper proposes convolutional neural networks for this task and presents an exhaustive comparison on latest LivDet 2011 and 2013 databases. Apart from classical classification nets, also metric-based deep siamese networks are evaluated learning a distance metric enforcing live-spoof pairs to be of higher distance than live-live pairs. This is useful for attended enrollment scenarios where a live gallery image is available (e.g. trusted-source fingerprint reference on the passport chip). Experiments reveal remarkable accuracy for all Convolutional Neural Networks (CNNs) CaffeNet (96.5%), GoogLeNet (96.6%), Siamese (93.1%), good material robustness (max. 5.6% diff.) but weak sensor-interoperability.

Liveness-based fusion approaches in multibiometrics

Recent works have shown that multimodal biometric systems are vulnerable to direct spoof attacks, even partial ones. On the other hand, liveness detection approaches have been proposed as a countermeasure. The aim of this paper is to devise proper approaches to integrate liveness detection into fusion at score or decision level, and then test them to assess which benefit can be obtained by smartly using vitality information in multibiometric systems.