The focus of Morphing Attack Detection (MAD) is to identify unauthorised attempts to use a legitimate identity. One common scenario involves creating altered images and using them in passport applications. Currently, there are limited datasets available for training the MAD algorithm due to privacy concerns and the challenges of obtaining and processing a large number of printed and scanned images. A larger and more diverse dataset representing passport application scenarios, including various devices and resulting printed, scanned, or compressed images, is needed to enhance the detection capabilities and identify such morphing attacks. However, generating training data that accurately represents the variety of attacks is a labour-intensive task since the training material is created manually. This paper presents two methods based on texture transfer techniques for the automatic generation of digital print and scan facial images, which are utilized to train a Morphing Attack Detection algorithm. Our proposed methods achieve an Equal Error Rate (EER) of 3.84% and 1.92% on the FRGC/FERET database when incorporating our synthetic and texture-transferred print/scan images at 600 dpi alongside handcrafted images, respectively.
