Morphing Attack Detection using Laplace operator based features

Authors

  • Ulrich Scherhag
  • Daniel Fischer
  • Sergey Isadskiy
  • Jonas Otte
  • Christoph Busch

Abstract

The vulnerability of facial recognition systems through morphing attacks is a known problem. Since the rst publication about this vulnerability of facial recognition systems, a variety of morphing attack detection methods have been presented, promising an automated detection of such fraudulent attacks. In this work, a new approach is presented attempting to distinguish bona de from morphed images based on information about the edges in the image extracted by the Laplace operator. It can be demonstrated that the features employed contain information that can contribute to the detection of morphed face images.

Published

2020-11-23