Morphing Attack Detection for Low-Power Electronics using the Residual Noise of Spatial and Frequency-based Features

摘要

Recently morphing attack becomes a serious challengein the travel documents as one of main systems in smart urbanmobility initiatives. Aiming at face morphing detection whichsuitable for low-power IoT devices, we propose morphing attackdetection using low-level features from residual noise in spatialand frequency domain. The result of the proposed method on theLabeled Faces in the Wild dataset achieve D-EER at 6.69%.