Fraunhofer scientists have developed a new machine learning, face recognition system to prevent face morphing.
Automated face recognition is widely used today for unlocking smartphones and speeding up airport security checks.
Yet they are also prone to being manipulated by hackers who adeptly meld two different facial images into one.
According to the scientists, a single passport featuring a photograph manipulated this way can be used by two different people simultaneously.
“Criminals are capable of tricking face recognition systems – like the ones used at border control – in such a way that two people can use one and the same passport,” explains Fraunhofer IPK scientist Lukasz Wandzik.
Wandzik and his colleagues at Fraunhofer HHI are part of the ANANAS project that’s developing a new process to identify image anomalies which occur during digital image processing in morphing processes.
ANANAS is a German acronym for ‘Anomaly Detection for Prevention of Attacks on Authentication Systems Based on Facial Images.’
Wandzik and his team have been applying modern image processing and machine learning methods such as deep neural networks for processing image data.
The project partners started by generating the data used to train the image processing programs to detect manipulations. Here different faces were morphed into one face.
“Using morphed and real facial images, we’ve trained deep neural networks to decide whether a given facial image is authentic or the product of a morphing algorithm,” explains Professor Peter Eisert, head of the Vision & Imaging Technologies department at Fraunhofer HHI.
“The networks can recognize manipulated images based on the changes occurring during manipulation, especially in semantic areas such as facial characteristics or reflections in the eyes.”
According to the scientists, the neural networks make very reliable decisions on whether or not an image is genuine, with an accuracy rate of over 90% in the test databases created in the project.
Image and content: Fraunhofer HHI