Differential Newborn Face Morphing Attack Detection using Wavelet Scatter Network

Face Recognition System (FRS) are shown to be vulnerable to morphed images of newborns. Detecting morphing attacks stemming from face images of newborn is important to avoid unwanted consequences, both for security and society. In this paper, we present a new reference-based/Differential Morphing Attack Detection (MAD) method to detect newborn morphing images using Wavelet Scattering Network (WSN). We propose a two-layer WSN with 250 × 250 pixels and six rotations of wavelets per layer, resulting in 577 paths. The proposed approach is validated on a dataset of 852 bona fide images and 2460 morphing images constructed using face images of 42 unique newborns. The obtained results indicate a gain of over 10% in detection accuracy over other existing D-MAD techniques.

I. Introduction

Morphing attacks using face images have demonstrated high degree of threat to automatic Face Recognition Systems (FRS). Morphing is the process of blending two or more face images, resulting in a constituent face image that visually resembles source face images used for morphing. The widespread availability of open source tools for morphing [1] –[3] further amplifies the threat, as it can facilitate morphing attack generation without the need for expert knowledge. It is well demonstrated in [4] that morphed images can successfully deceive human observers, including trained experts like border guards and ID experts. Threats on FRS coupled with weakness of human observers in detecting morphed images intensify the attack strength on real-life applications including border control and remote ID verification. Morphing detection of newborn faces therefore becomes critical to avoid various problems such as illegal adoption, sexual exploitation, child marriages, and organ harvesting.