Face Morph Detection: A Systematic Review

Abstract

Biometric verification especially faces recognition has been widely applied in daily life, including in immigration. Face recognition is used for identification verification in passport identities. But there is a security threat in face recognition, which is called face morph. Face morph is the combination of the passport applicant's face image with the face image of another subject to produce a face image similar to the face of the applicant's passport document by covering the facial features of the other subject. It can cause serious problems if misused by people listed on the blacklist. Blacklists are people who are subject to immigration restrictions such as suspension of passport issuance, inclusion on the prevention list, and inclusion on the deterrence list. So that the person listed on the blacklist can have a passport to then cross the border out of Indonesian territory which can cause problems and disturb the country's sovereignty. This research uses the Systematic Literature Review (SLR) method to collect and analyze previous studies related to the research topic. The result of this research is to get the method and trend of face morph detection. Detection of face morphs is generally divided into 2 scenarios, i.e., single image-based and differential image-based. The detection method for single image-based is divided into texture, quality, noise, deep learning, and hybrid, while the detection method for differential image-based is divided into feature difference and demorphing. So, with face morph detection, it is expected to be able to increase the security of passport documents and increase security at immigration checkpoints.