da/sec scientific talk on Biometrics

Topic: Quality Degradation Impact on Differential Morphing Attack Detection

by Torsten Schlett
D19/2.03a, May 15, 2024 (Wednesday), 12.00 noon

Keywords — Face Recognition, Morphing Attack Detection, Quality Assessment

Abstract

„Morphed face images combine two or more images from different individuals. In a border control scenario, a passport with a morphed face image can be used to deceive both human examiners and automated face recognition systems.
Differential Morphing Attack Detection (D-MAD) attempts to detect morphed images through a comparison against a trusted face image, such as a live capture for border control. While the quality of a suspected passport image may be expected to be checked as part of the passport application process, the quality of a trusted live capture might be more strongly affected by environmental factors. Face image quality assessment (FIQA) algorithms could however be used to automatically detect such cases.
This talk presents an analysis of the D-MAD performance impact and FIQA response for such cases, based on a set of synthetic degradation configurations.“