Simulating Print/Scan Textures for Morphing Attack Detection
Juan E. Tapia, Christoph Busch, Haoyu Zhang, Raghavendra Ramachandra, Kiran Bylappa Raja · 2023
Morphing Attack Detection is a relevant topic aiming to detect attempts of unauthorised individuals who want access to a “valid” identity. One of the main scenarios is printing morphed images and submitting the respective print in a passport application process. In order to improve the detection capabilities and spot such morphing attacks, it will be necessary to have a more realistic data set representing the passport application scenario with the diversity of devices and the resulting printed scanned or compressed images. Creating training data representing the diversity of attacks is a very demanding task because the training material is developed manually or semiautomatically. This paper proposes a transfer-style pixel-wise network for a general-purpose method to automatically create digital print/scan face images and use such images in the training of a Morphing Attack Detection (MAD) method. Our proposal can reach an Equal Error Rate (EER) of 5.13% with Random Forest and 3.17% using MobileNetV2 on the FRGCv2 database between manual print/scan and synthetics print/scan with 600 dpi. This method opens a new insight into developing attack datasets easily and time efficiently.