Event Details

Dr Sophie Nightingale, Lecturer in Psychology 9th June at 12.30

Title: Perceptual and computational approaches for detecting face morphing

Description: We frequently rely on photo-based documents for identity verification in critical settings such as border control. Yet an extensive body of research shows that matching pairs of unfamiliar faces is a difficult task, even for trained identification-checkers. The difficulty of this task leaves identity verification processes vulnerable to fraudulent attacks. A relatively new type of fraud that border control agencies are facing is the use of morphed passport photos—morphing involves digitally combining images of two individuals to create a new image that maintains a likeness to each of the original identities. Early research exploring human detection of morphed images indicates that people frequently accept both low-quality and high-quality morphed images as genuine. The face databases used to create the stimuli in these previous studies, however, have a number of limitations, namely, that the morphs are low-quality and/or limited in terms of race, gender, and age diversity. The current research extends this previous literature by examining human and computer-based detection of face morphing using a diverse set of facial images to generate high-quality morphs.

All Welcome-Teams link here