Roland Rauner defended his Bachelor Thesis on ‘Analyse der Kombination von verschiedenen Einflussfaktoren auf die differenzielle Morphing-Angriffs Detektion’
The differential morphing attack detection is intended to detect the manipulation by comparing the extracted features from the passport photo and an additional live recording. Since this is a relatively young research area, there is still a lot of potential in terms of detection rates, the analysis of all influencing factors as well as the generalization ability through various morphing methods and databases.
Thus, the focus of this work is the precise analysis of these priorities. With current morphing tools, standardized metrics as well as different databases, the generalization ability of various biometric feature extraction algorithms is examined and a statement about their suitability for the detection of morphing attacks at the automated border control is formulated.