Artificial Image Tampering Distorts Spatial Distribution of Texture Landmarks and Quality Characteristics

View on arXiv ← Back to list

Authors: Tahir Hassan, Aras Asaad, Dashti Ali, Sabah Jassim

Published: 2022-08-04 15:13:00+00:00

AI Summary

This paper proposes two novel feature vectors for detecting morphed face images: a multi-characteristic image quality feature (MCIQ) and features derived from persistent homology (PH) of texture landmarks. These features, combined with an SVM classifier, achieve low error rates in detecting morphed faces, suitable for constrained devices.

Abstract

Advances in AI based computer vision has led to a significant growth in synthetic image generation and artificial image tampering with serious implications for unethical exploitations that undermine person identification and could make render AI predictions less explainable.Morphing, Deepfake and other artificial generation of face photographs undermine the reliability of face biometrics authentication using different electronic ID documents.Morphed face photographs on e-passports can fool automated border control systems and human guards.This paper extends our previous work on using the persistent homology (PH) of texture landmarks to detect morphing attacks.We demonstrate that artificial image tampering distorts the spatial distribution of texture landmarks (i.e. their PH) as well as that of a set of image quality characteristics.We shall demonstrate that the tamper caused distortion of these two slim feature vectors provide significant potentials for building explainable (Handcrafted) tamper detectors with low error rates and suitable for implementation on constrained devices.

Key findings

The proposed MCIQ and PH-based features achieve low false rejection rates (FRR) and false acceptance rates (FAR) in detecting morphed faces across two datasets. The model generalizes well across datasets, achieving less than 4% FRR and FAR in cross-database validation. The low dimensionality of the features makes them suitable for implementation on resource-constrained devices.

Approach

The approach uses two feature vectors: MCIQ, a reference-less spatio-statistical image quality feature vector capturing the scatter of various quality characteristics; and PH features extracted from texture landmarks (ULBP) to quantify spatial distortions caused by morphing. These features are then used with a Support Vector Machine (SVM) classifier for detection.

Datasets

AMSL DB and Utrecht DB

Model(s)

Support Vector Machine (SVM) with cubic kernel

Author countries

UK, Canada

← Previous