Fingerprinting and forensic analysis of multimedia

  • Authors:
  • Daniel Schonberg;Darko Kirovski

  • Affiliations:
  • University of California, Berkeley, CA;Microsoft Research, Redmond, WA

  • Venue:
  • Proceedings of the 12th annual ACM international conference on Multimedia
  • Year:
  • 2004

Quantified Score

Hi-index 0.00

Visualization

Abstract

One of the prime reasons movie and music studios have ignored the Internet for open-networked multimedia content delivery, has been the lack of a technology that can support a secure digital rights management (DRM) system on a general purpose computer. The difficulty of building an effective multimedia DRM stems from the fact that traditional cryptograic primitives such as encryption or scrambling do not protect audio or video signals once they are played in plain-text. This fact, commonly referred to as "the analog hole," has been responsible for the popularity of multimedia file sharing which cannot be controlled, at least technically, by content's copyright owners. In this paper, we explore multimedia fingerprinting as an answer to "the analog hole" problem. We propose a new ase-shifted spread-spectrum fingerprinting paradigm particularly tailored for fast detection. Next, we present two techniques for fast maximum-likelihood audio and video synchronization designed to cope with typical de-synch attacks. We analyze the collusion resistance of a large class of spread-spectrum fingerprinting systems using a new, gradient attack. Surprisingly, we show that the collusion resistance of traditional spread-spectrum fingerprints is a small constant that does not depend on the object size.