Presentation attack detection (PAD) is the set of techniques that determine whether a biometric sample is presented by a genuine, live subject or by an artefact intended to spoof the system. Attacks include printed photographs, replayed video, silicone fingerprints, and three-dimensional masks. PAD, also known as liveness or anti-spoofing analysis, is standardised under ISO/IEC 30107 and is essential to the trustworthiness of biometric authentication.
Overview
- A biometric system is only as strong as its weakest enrolment path, and presentation attacks exploit the gap between capturing a likeness and confirming a living person. PAD methods analyse texture, motion, depth, physiological signals, and challenge-response behaviour to separate genuine presentations from artefacts. They are typically evaluated using the standardised metrics of attack presentation classification error rate and bona fide presentation classification error rate.
Key aspects
- Passive liveness from texture, micro-movement, and reflectance cues
- Active challenge-response such as blink, smile, or head-turn prompts
- Depth and infrared sensing to defeat flat-image spoofs
- Physiological signals including remote photoplethysmography
- Standardised evaluation under ISO/IEC 30107 (APCER, BPCER)
Applications
- Face unlock on mobile devices and laptops
- Remote identity verification for onboarding and KYC
- Border control and automated passport gates
- Fingerprint and iris systems in high-assurance access control