Many countries (including Germany, since November 1, 2010) now require the use of biometric images in e-passports and other ID documents, such as driver’s licenses, to guarantee accurate identification.
But not all biometric samples are equally suitable for the automated recognition of an individual person. Low-quality photos frequently lead to errors—a weakness that criminals exploit through face morphing, i.e., when two faces are combined. ATHENE researchers at Fraunhofer IGD are therefore looking into methods of assessing the quality of facial images.
To date, the requirements specified for biometric ID photos are limited to how they are taken—i.e., the background, the facial expression, the pose, and permissible accessories. In the future, the goal is to expand specifications to include quality attributes: Metrics will be employed to identify images that are unsuitable for automatic facial recognition.
If methods of assessing the quality of facial images are successfully developed, they could be advantageous in a number of use cases—including automated border control systems, such as the proposed entry/exit system for the Schengen Area.