Fingerprint Presentation Attack Detection Based on Local Features Encoding for Unknown Attacks

Fingerprint-based biometric systems have experienced a large development in the past. In spite of many advantages, they are still vulnerable to attack presentations (APs). Therefore, task determining whether sample stems from live subject (i.e., bona fide) or an artificial replica is mandatory requirement which has recently received considerable attention. Nowadays, when materials for fabrication Presentation Attack Instruments (PAIs) been used train Detection (PAD) methods, PAIs can be successfully identified most cases. However, current PAD methods face difficulties detecting built unknown and/or recepies, acquired using different capture devices. To tackle this issue, we propose new technique based on three image representation approaches combining local and global information fingerprint. By transforming these representations into common feature space, correctly discriminate fide aforementioned scenarios. The experimental evaluation our proposal over LivDet 2011 2019 databases, yielded error rates outperforming top state-of-the-art results by up 72% challenging addition, best achieved competition (overall accuracy 96.17%).