Robust Fingerprint Spoof Identifier

Abstract

Fingerprint is commonly used method for biometric identification. Nowadays, every person is having a smartphone and it has a inbuilt fingerprint scanner for authentication purpose. There are many applications which use fingerprint scanner for authentication purpose such as banking applications, WhatsApp etc. If fingerprint gets spoofed then the system may get vulnerable, user's personal data can get exposed. It is very crucial to design a system which is reliable and robust for detecting spoofing attacks. Specially, we are using transfer learning in deep CNN with enhanced fingerprint images. The dataset that we are using is from public domain Liv-det dataset. The dataset is from year 2011, 2013, 2015 respectively. We also provide a graphical user interface that shows whether the fingerprint is live or spoof along with the spoofness score