Fingerprint Spoof Buster (1712.04489v1)

Abstract: The primary purpose of a fingerprint recognition system is to ensure a reliable and accurate user authentication, but the security of the recognition system itself can be jeopardized by spoof attacks. This study addresses the problem of developing accurate, generalizable, and efficient algorithms for detecting fingerprint spoof attacks. Specifically, we propose a deep convolutional neural network based approach utilizing local patches centered and aligned using fingerprint minutiae. Experimental results on three public-domain LivDet datasets (2011, 2013, and 2015) show that the proposed approach provides state-of-the-art accuracies in fingerprint spoof detection for intra-sensor, cross-material, cross-sensor, as well as cross-dataset testing scenarios. For example, in LivDet 2015, the proposed approach achieves 99.03% average accuracy over all sensors compared to 95.51% achieved by the LivDet 2015 competition winners. Additionally, two new fingerprint presentation attack datasets containing more than 20,000 images, using two different fingerprint readers, and over 12 different spoof fabrication materials are collected. We also present a graphical user interface, called Fingerprint Spoof Buster, that allows the operator to visually examine the local regions of the fingerprint highlighted as live or spoof, instead of relying on only a single score as output by the traditional approaches.

• The paper presents a deep CNN method using local minutiae-centered patches for robust fingerprint spoof detection.
• The method achieves state-of-the-art accuracy on LivDet datasets, significantly outperforming prior techniques (e.g., 99.03% vs 95.51% on LivDet 2015).
• This approach enhances security, scalability, and adaptability for fingerprint authentication systems, promising broad commercial and security applications.

Fingerprint Spoof Buster: Liveness Detection using Local Patches

This paper presents a method for fingerprint spoof detection using local patches, centered and aligned based on fingerprint minutiae, processed through a deep convolutional neural network (CNN). This approach is pivotal because fingerprint recognition systems are increasingly vulnerable to spoof attacks through presented biometric data. The technique achieves state-of-the-art performance in detecting spoof fingerprints across various datasets and sensor types.

Key Findings and Methodology

The paper emphasizes the need for robust spoof detection algorithms due to increased application of fingerprint recognition systems and their susceptibility to presentation attacks. Presentation attacks are defined as attempts to deceive biometric systems using artificial materials like gummy fingers, printed fingerprint targets, or altered and cadaver fingers. The paper proposes a CNN-based approach utilizing local patches extracted around fingerprint minutiae to differentiate between live and spoof fingerprints.

Experimental Validation:

• The effectiveness of this approach is validated through extensive experiments on LivDet datasets from 2011, 2013, and 2015. The proposed method surpasses existing solutions with impressive accuracies across several testing scenarios including intra-sensor, cross-material, and cross-dataset evaluations.
• For instance, using the LivDet 2015 dataset, the algorithm achieved 99.03% average accuracy compared to 95.51% by prior competition winners. This significant performance boost can be attributed to the use of local minutiae-centered patches rather than whole fingerprint images.

Datasets and Implementation:

• Two new datasets were collected, comprising over 20,000 fingerprint images from different sensors and spoof materials, enhancing the robustness of validation efforts.
• A graphical user interface, "Fingerprint Spoof Buster," has been developed, allowing examiners to review local fingerprint regions identified as either live or spoof.

Theoretical Implications

The method's utilization of local patches forms a comprehensive analysis regimen, improving the detection of spoof fingerprints under various conditions. This approach shifts from traditional whole-image processing to region-specific analysis, which enhances the generalization of spoof detection across new materials and unobserved alterations. The minute-level detail captured by local patching paves the way for future methodologies in biometric authentication systems.

Practical Implications and Future Directions

The proposed system significantly elevates the security layer in fingerprint-based applications. Its scalability and adaptability in processing and real-time evaluation open pathways for integrating such systems in commercial, governmental, and personal security domains. Moreover, with the advent of smartphones and security applications everywhere, the approach holds promise for broad commercial appeal due to its high accuracy and low processing time.

Future developments might involve integrating the patch-based approach with newer deep learning architectures like transformers, refining cross-material detection, and addressing emerging spoofing technologies. Continuous improvement of such systems will require adapting to novel presentation attack instruments as they appear, ensuring the robustness and reliability of biometric authentication systems effectively withstands any foreseeable threats.