A Survey on Deep Learning Techniques for Fingerprint Presentation Attack Detection

TL;DR

This paper reviews deep learning methods for detecting fake fingerprints in security systems.

Contribution

The paper provides a comprehensive survey and categorization of deep learning-based fingerprint presentation attack detection techniques.

Findings

Deep learning-based FPAD methods outperform traditional handcrafted approaches in reliability and generalization.

The paper categorizes recent FPAD methods and discusses benchmark metrics and datasets.

Future research directions are outlined to improve the field of FPAD.

Abstract

The vulnerabilities of the fingerprint authentication system have raised security concerns in terms of adapting them in highly secured access control applications. Therefore, fingerprint presentation attack detection (FPAD) methods are essential to ensure reliable fingerprint authentication. Due to the lack of generalization of the traditional handcrafted-based approaches, deep learning-based FPAD has become mainstream and achieves remarkable performance in the past decade. In this paper, we will concentrate only on deep learning-based FPAD methods. We investigate recent methods and divide those into different categories to provide a comprehensive description. The benchmark metrics and publicly available datasets are also discussed. Lastly, we conclude the paper by discussing future perspectives to inspire further research.

Figures