On the Human Ability in Detecting Digitally Manipulated Face Images

Abstract

Facial image manipulation in the context of electronic ID documents is a concrete security threat, confirmed by several real cases reported by the authorities of different countries. Such manipulations have a negative impact on the automated face recognition accuracy and should therefore be identified before the altered image is included in the document. This study reports and discusses the result of a test carried out on human examiners to evaluate their ability in detecting digital manipulations of facial images.