.png){kind=icon}
.png){kind=icon}
Return
Face morphing attack detection and attacker identification based on a watchlist
.png){kind=icon}
PRE
AI
DOI：10.1016/j.image.2022.116748.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Face verification is widely used for Automated Border Control (ABC) in many countries. But such ABC systems are vulnerable to Morphing Attacks (MAs), where a morphed face image is submitted to apply for a passport. To counter face MAs, this paper proposes to detect morphed face images and identify morphing attackers by the use of a watchlist. It is carried out by comparing a suspect image with the biometric references contained in a watchlist, and its detection process is accomplished by analyzing the results of face comparison. Once a morphed image is detected, its morphing attacker is also identified. Meanwhile, a database with different morphing methods, image qualities, facial expressions, and face angles is collected. Experimental results and analysis show that it can achieve stable detection and attacker identification performance for 4 different face MAs, and it can well generalize to unseen morphing types and weights. Moreover, it has good robustness to variations of image qualities, facial expressions, and face angles.
Keywords:
Automated border control
Face verification
Morphing attacks
Morphing attack detection
Watchlist
Journal
S
IF:
2.7
Papers:
2.8K
Citations:
4.2K