Watermark-based authentication

The large variety of signal processing tools available today permits to anyone to easily modify digital multimedia objects like audio, still images or video without leaving any perceptible trace of the modifications. Even though these tools are extremely useful in several scenarios, e.g. to increase the perceived quality of multimedia contents, there are cases in which we would like to be sure that a digital content is authentic, i.e. it corresponds to the original version. Stated in another way, we would like to find a way to avoid the loss of credibility of digital data. Authentication is the science that studies how to protect the integrity of digital media. Generally speaking, there are two basic forms of authentication: passive and active authentication. Passive authentication, also called forensic analysis, tries to understand whether a digital content has been tampered with by using statistical analysis without previously adding an authentication signal to the digital media. Passive authentication has the desirable characteristic of working with virtually any type of data, without requiring that they are modified at the time of creation. On the negative side, passive authentication is not always possible and doubts exist about its reliability and security. This is not the case with active authentication, whereby the integrity of a digital content is protected (and demonstrated) by embedding an authenticating signal within the digital content itself before sharing it with other users. Active authentication is also called watermark-based authentication, since it uses watermarking technology to embed the authenticating signal within the to-be-protected support. © 2010 Springer-Verlag Berlin Heidelberg.