Assessing Transfer Learning on Convolutional Neural Networks for Patch-Based Fingerprint Liveness Detection
Fingerprint based biometric identification systems are vulnerable to spoofing attacks that involve the use of fake replicas of real fingerprints. The resulting security issues can be mitigated through the development of software modules capable of detecti
- PDF / 823,402 Bytes
- 17 Pages / 439.37 x 666.142 pts Page_size
- 71 Downloads / 227 Views
Abstract Fingerprint based biometric identification systems are vulnerable to spoofing attacks that involve the use of fake replicas of real fingerprints. The resulting security issues can be mitigated through the development of software modules capable of detecting the liveness of an input image and, thus, of discarding fake fingerprints before the classification step. In this work we present a fingerprint liveness detection method that combines a patch-based voting approach with Transfer Learning techniques. Fingerprint images are first segmented to discard background information. Then, small-sized foreground patches are extracted and processed by popular Convolutional Neural Network models, whose pre-trained versions were adapted to the problem at hand. Finally, the individual patch scores are combined to obtain the fingerprint label. Experimental results on well-established benchmarks show the promising performance of the proposed method compared with several state-of-the-art algorithms. Keywords Fingerprint spoofing · Convolutional Neural Networks · Fingerprint segmentation · Patch based classification · Transfer learning
1 Introduction In recent years, fingerprint based authentication has become more and more pervasive [1]. Fingerprint sensors are being deployed on a variety of consumer devices, like A. Toosi · S. Cumani · A. Bottino (B) Department of Control and Computer Engineering, Politecnico di Torino, Turin, Italy e-mail: [email protected] A. Toosi e-mail: [email protected] S. Cumani e-mail: [email protected] © Springer Nature Switzerland AG 2019 C. Sabourin et al. (eds.), Computational Intelligence, Studies in Computational Intelligence 829, https://doi.org/10.1007/978-3-030-16469-0_14
263
264
A. Toosi et al.
notebooks and mobile phones, and are becoming a solution for access control in common facilities, like schools, health clubs and hospitals. The wide adoption of these kind of devices, however, rises several security concerns, due to their vulnerability to different forms of attack, which might result in granting access to unauthorized persons. The development of effective countermeasures has therefore become a relevant topic. Attacks can be divided into two categories: direct attacks, operating directly on the sensor, usually by means of fake replicas of real fingerprints, and indirect attacks, which target the inner modules of the fingerprint recognition system. Clearly, the first kind of attacks are easier to implement for intruders without expert knowledge of the hardware and software architectures of the sensors. Fingerprint replicas can be easily obtained by creating a mold from a latent or real fingerprint, and then filling it with materials like latex, gelatin, vinyl or wood glue and so on. It has been demonstrated that even a high quality digital image of a fingerprint allows performing successful attacks [2]. The literature shows that the success rate of such spoofing attacks can be higher than 70% [3], highlighting the need for specific protection methods capa
Data Loading...
