Elsevier

Pattern Recognition

Volume 48, Issue 4, April 2015, Pages 1050-1058
Pattern Recognition

Local contrast phase descriptor for fingerprint liveness detection

https://doi.org/10.1016/j.patcog.2014.05.021Get rights and content

Highlights

  • We propose a new local descriptor for fingerprint liveness detection.

  • It is based on the joint use of contrast and phase information.

  • Image analysis is carried out in both the spatial and the transform domains.

  • We generate a bi-dimensional contrast–phase histogram, used as feature vector.

  • A properly trained linear-kernel SVM classifier makes the final live/fake decision.

Abstract

We propose a new local descriptor for fingerprint liveness detection. The input image is analyzed both in the spatial and in the frequency domain, in order to extract information on the local amplitude contrast, and on the local behavior of the image, synthesized by considering the phase of some selected transform coefficients. These two pieces of information are used to generate a bi-dimensional contrast-phase histogram, used as feature vector associated with the image. After an appropriate feature selection, a trained linear-kernel SVM classifier makes the final live/fake decision. Experiments on the publicly available LivDet 2011 database, comprising datasets collected from various sensors, prove the proposed method to outperform the state-of-the-art liveness detection techniques.

Introduction

Biometric systems are commonly used for authentication in various security applications. By relying on features that are unique for each individual (iris, fingerprints, etc.) they guarantee simplicity and reliability at the same time, avoiding the typical problems of systems based on the use of passwords, which can be forgotten, stolen, or figured out. Of course, biometric systems have their own weaknesses. In particular, they are relatively vulnerable to some sophisticated forms of spoofing. Fingerprint-based systems are among the most commonly used and, for this very same reason, more subject to attacks. Indeed, early systems could be easily fooled by fake fingerprints, reproduced on simple molds made of materials such as silicone, Play-Doh, clay or gelatin [1], [2].

A large number of methods have been proposed in recent years to combat spoofing. Some of them rely on dedicated additional hardware embedded in the sensor which confirms the vitality of the fingerprint by measuring temperature, odor, pulse oxiometry, etc. [3], [4]. Since this approach can use different sources of information, it turns out to be more resilient to generic attacks, and typically guarantees a very good reliability. Nonetheless, this is a relatively expensive and rigid solution, and for these reasons has gained little popularity.

Software-based methods, based on signal-processing techniques, are certainly more appealing, for their reduced cost and invasiveness, and their higher flexibility. They try to detect vitality by analyzing synthetic image features that are peculiar of vital fingerprints and not easily reproduced on fakes. Such features are typically singled out based on a deep study of the physics of the problem and/or a careful analysis of the statistical behaviour of the captured images. A large number of such methods have been proposed in recent years [5], [6], [7], [8], [9], [10], [11], [12], [13], [14], [15], [16], [17], [18], [19], based on clever and well-founded ideas, testifying on the relevance and difficulty of this problem.

In parallel with these approaches based on a global description of fingerprints, though, techniques based on machine learning and local descriptors have been taking hold recently for fingerprint liveness detection [20], [21], [22], [23], [24]. Local descriptors, as the name suggests, describe the statistical behavior observed locally in very small patches of the image by means of histograms (frequencies of occurrence, empirical probability distributions) collected over the ensemble of all patches. These histograms are then used as features to classify the images by means of conventional classification approaches. Techniques based on local descriptors provide usually a much better performance than the previous class of methods. It is somewhat surprising that these alternative methods, showing often little or no fingerprint-specific clues, overcome the global-descriptor approaches. On the other hand, liveness detection, as all tasks related with digital security, can be seen as a game with two players, and it is only reasonable to expect that methods based on coarse features will be sooner or later tricked by smart attackers equipped with better materials and better knowledge of the fingerprint statistics. Local descriptors represent a natural evolution towards the discovery of fine features that are more discriminating and also harder to tamper with.

Till now, however, only general-purpose descriptors have been considered for this task, while descriptors conceived specifically for fingerprint images could very likely provide a still better performance. In this work, we move some steps towards this goal. Building upon previous work in this field, we propose a new local descriptor based on spatial-domain and transform-domain features which appear to be very discriminating for liveness detection. An appreciable performance gain w.r.t. the state of the art can be observed on standard and publicly available datasets, which is an encouragement to further pursue this line of work.

The rest of the paper is organized as follows: in Section 2 we briefly review the literature on fingerprint liveness detection, with special focus on techniques based on local descriptors, then in Section 3 describe the proposed approach, and in Section 4 present the results of some numerical experiments. Eventually, in Section 5, we draw conclusions and outline future researches.

Section snippets

Conventional methods

Because of imperfections in the material used, spoofed fingerprints exhibit usually a worse quality than the live ones. Based on this consideration in [6] the coarseness of the fingerprint is used as a discriminative feature. After subtracting a denoised version of the image from the original one, the variance of the noise residual in detail wavelet subbands is considered as a measure of coarseness and used for classification. Also in [18] quality-related measures are used for fingerprint

Proposed local descriptor

While in [24] we just concatenated features, summing up their dimensionality, in this work we propose a more clever combination, following also [46], [50]. Eventually, our proposed descriptor is based on a spatial-domain component, inspired to the homologous component of WLD, and a phase component, which is the rotation-invariant version of LPQ, ending up with the Local Contrast-Phase Descriptor (LCPD) illustrated below. In the next three subsections we provide details on the two components and

Experimental results

In this Section, after presenting the standard datasets used for evaluating the performance of our liveness detection method, we describe the procedure employed for setting up its parameters and validate experimentally the main design choices. Finally, we compare performance with several state-of-the-art techniques.

Conclusions

In the last few years, techniques based on local descriptors have provided a significant performance leap in fingerprint liveness detection. Till now, however, only general-purpose descriptors have been considered for this task, while descriptors conceived specifically for fingerprint images could be expected to provide a still better performance. In this work, we moved some steps towards this goal, analyzing a number of features, their possible variants, and their combinations, for the

Conflict of interest statement

None declared.

Diego Gragnaniello received the Laurea degree in computer engineering from the University Federico II of Naples, Italy, in 2011. He is currently a Ph.D. student in electronic and telecommunications engineering at the University of Naples. His study and research interests include image processing, particularly liveness detection and forgery detection.

References (55)

  • F. Liu et al.

    WLBPweber local binary pattern for local image description

    Neurocomputing

    (2013)
  • T. Matsumoto, H. Matsumoto, K.Yamada, S.Hoshino, Impact of artificial gummy fingers on fingerprint systems, in:...
  • D. Baldissera, A. Franco, D. Maio, D. Maltoni, Fake fingerprint detection by odor analysis, in: International...
  • P. Reddy et al.

    A new antispoofing approach for biometric devices

    IEEE Trans. Biomed. Circuits Syst.

    (2008)
  • Y. Moon et al.

    Wavelet based fingerprint liveness detection

    Electron. Lett.

    (2005)
  • S. Parthasaradhi et al.

    Time-series detection of perspiration as a liveness test in fingerprint devices

    IEEE Trans. Syst. Man Cybern. C Appl. Rev.

    (2005)
  • B. Tan, S. Schuckers, Liveness detection for fingerprint scanners based on the statistics of wavelet signal processing,...
  • A. Antonelli et al.

    Fake finger detection by skin distortion analysis

    IEEE Trans. Inf. Forensics Secur.

    (2006)
  • C. Jin, H. Kim, S. Elliott, Liveness detection of fingerprint based on band-selective fourier spectrum, in: Proceedings...
  • P. Coli, G. Marcialis, F. Roli, Power spectrum-based fingerprint vitality detection, in: Proceedings of IEEE Workshop...
  • S. Nikam, S. Agarwal, Gabor filter-based fingerprint anti-spoofing, in: Lecture Notes in Computer Science, vol. 5259,...
  • B. Tan et al.

    New approach for liveness detection in fingerprint scanners based on valley noise analysis

    J. Electron. Imaging

    (2008)
  • G. Marcialis, F. Roli, A.Tidu, Analysis of fingerprint pores for vitality detection, in: International Conference on...
  • J. Galbally et al.

    Image quality assessment for fake biometric detectionapplication to iris, fingerprint and face recognition

    IEEE Trans. Image Process.

    (2014)
  • S. Nikam, S. Agarwal, Texture and wavelet-based spoof fingerprint detection for fingerprint biometric systems, in:...
  • L. Ghiani, G. Marcialis, F.Roli, Fingerprint liveness detection by local phase quantization, in: International...
  • L. Ghiani, A. Hadid, G. Marcialis, F. Roli, Fingerprint liveness detection using binarized statistical image features,...
  • Cited by (133)

    • A multilayer system to boost the robustness of fingerprint authentication against presentation attacks by fusion with heart-signal

      2022, Journal of King Saud University - Computer and Information Sciences
      Citation Excerpt :

      In the software-based approach, some liveness features are extracted from the fingerprint itself. Some of the frequently used features that have been investigated for fingerprint liveness detection are the local binary pattern (LBP) (Nikam and Agarwal, 2008), local phase quantization (LPQ) (Ghiani et al., 2012), weber local descriptor (WLD) (Gragnaniello et al., 2013), and local contrast phase descriptor (LCPD) (Gragnaniello et al., 2015). In these handcrafted techniques, expert knowledge is required for feature extraction and classification.

    • Fingerprint liveness detection through fusion of pores perspiration and texture features

      2022, Journal of King Saud University - Computer and Information Sciences
      Citation Excerpt :

      Since, the software based methods include only image processing algorithms and does not require extra hardware, they are most prevailing now-a-days. Number of algorithms in the literature that has presented the software based approach are mainly, skin deformation based (Abhyankar and Schuckers, 2004; Antonelli et al., 2006; Zhang et al., 2007; Tan and Schuckers, 2008), handcrafted texture features based (Champod et al., 2004; Mura et al., 2015; Tan and Schuckers, 2008; Moon et al., 2005; Abhyankar and Schuckers, 2006; Lee et al., 2009; Marasco and Sansone, 2010; Nikam and Agarwal, 2010; Jin et al., 2011; Galbally et al., 2012, 2013; Gottschlich et al., 2014; Gragnaniello et al., 2015; Ghiani et al., 2016; Agrawal et al., 2019; Sharma and Dey, 2019; Tan et al., 2020), based on texture features extracted from deep learning techniques (Nogueira et al., 2016; Jang et al., 2017; Yuan et al., 2017, 2019a, 2019b, 2020a, 2020b; Chugh et al., 2017; Zhang et al., 2019, 2020; Jung et al., 2019), and pore features based (Espinoza et al., 2011; Memon et al., 2011; Espinoza and Champod, 2011b; li et al., 2015; lu et al., 2015; Marcialis et al., 2010; Derakhshani et al., 2003; Schuckers and Abhyankar, 2004; Parthasaradhi et al., 2005; Tan and Schuckers, 2005, 2006a, 2006b, 2010; Abhyankar and Schuckers, 2008, 2010; DeCann et al., 2009; Johnson and Schuckers, 2014). Due to the fact that pore-based automated fingerprint recognition systems are efficient and achieves high matching accuracy, recently, pores based fingerprint liveness detection approach become the most explorable among the researchers.

    View all citing articles on Scopus

    Diego Gragnaniello received the Laurea degree in computer engineering from the University Federico II of Naples, Italy, in 2011. He is currently a Ph.D. student in electronic and telecommunications engineering at the University of Naples. His study and research interests include image processing, particularly liveness detection and forgery detection.

    Giovanni Poggi received the Laurea degree in electronic engineering from the University Federico II of Naples, Italy, in 1988. He is currently a Professor of telecommunications with the University Federico II of Naples, Italy, Department of Electrical Engineering and Information Technology and Coordinator of the Telecommunication Engineering School. His current research interests are focused on statistical image processing, including compression, restoration, and segmentation of remote-sensing images, both optical and SAR, and digital image forensics. Prof. Poggi has been an Associate Editor for the IEEE Transactions on Image Processing and Elsevier Signal Processing.

    Carlo Sansone is currently a Full Professor of Computer Science at the Department of Electrical Engineering and Information Technology of the University of Naples Federico II. His research interests cover the areas of image analysis and recognition, pattern recognition, graph matching and information fusion. Prof. Sansone has authored more than 150 research papers in international journals and conference proceedings. He is an Associate editor of Electronic Letters on Computer Vision and Image Analysis and Elsevier Information Fusion. He was also co-editor of two special issues on International Journals and three books. Prof. Sansone is a Vice-President of the GIRPR – the Italian Chapter of the International Association for Pattern Recognition (IAPR) and is a member of the IEEE.

    Luisa Verdoliva received the Laurea degree in telecommunication engineering and the Ph.D. degree in information engineering from the University Federico II of Naples, Italy, in 1998 and 2002, respectively. She is currently an Assistant Professor of telecommunications with the Department of Electrical Engineering and Information Technology, University of Naples Federico II. Her research is on image processing, in particular compression and restoration of remote-sensing images, both optical and SAR, and image digital forensics.

    View full text