Local contrast phase descriptor for fingerprint liveness detection
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)
- et al.
An evaluation of direct attacks using fake fingers generated from ISO templates
Pattern Recognit. Lett.
(2010) - et al.
Determination of vitality from a non-invasive biomedical measurement for use in fingerprint scanners
Pattern Recognit.
(2003) - et al.
Ridgelet-based fake fingerprint detection
Neurocomputing
(2009) - et al.
Spoofing protection for fingerprint scanner by fusing ridge signal and valley noise
Pattern Recognit.
(2010) - et al.
Combining perspiration- and morphology-based static features for fingerprint liveness detection
Pattern Recognit. Lett.
(2012) - et al.
A high performance fingerprint liveness detection method based on quality related features
Futur. Gener. Comput. Syst.
(2012) - et al.
Multi-scale local binary pattern with filters for spoof fingerprint detection
Inf. Sci.
(2014) - et al.
Copy-move forgery detection using multiresolution local binary patterns
Forensic Sci. Int.
(2013) - et al.
Local phase quantization for blur-insensitive image analysis
Image Vis. Comput.
(2012) - et al.
Local binary pattern (LBP) and local phase quantization (LPQ) based on Gabor filter for face representation
Neurocomputing
(2013)
WLBPweber local binary pattern for local image description
Neurocomputing
A new antispoofing approach for biometric devices
IEEE Trans. Biomed. Circuits Syst.
Wavelet based fingerprint liveness detection
Electron. Lett.
Time-series detection of perspiration as a liveness test in fingerprint devices
IEEE Trans. Syst. Man Cybern. C Appl. Rev.
Fake finger detection by skin distortion analysis
IEEE Trans. Inf. Forensics Secur.
New approach for liveness detection in fingerprint scanners based on valley noise analysis
J. Electron. Imaging
Image quality assessment for fake biometric detectionapplication to iris, fingerprint and face recognition
IEEE Trans. Image Process.
Cited by (133)
Deep learning techniques for biometric security: A systematic review of presentation attack detection systems
2024, Engineering Applications of Artificial IntelligenceA realtime fingerprint liveness detection method for fingerprint authentication systems
2023, Advances in ComputersA 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 SciencesCitation 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 SciencesCitation 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.
DEHFF – A hybrid approach based on distinctively encoded fingerprint features for live fingerprint detection
2022, Biomedical Signal Processing and Control
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.