Multimedia Forensics

Multimedia Forensics includes a set of scientific techniques recently proposed for the analysis of multimedia signals (audio, videos, images) in order to recover probative evidences from them; in particular, such technologies aim to reveal the history of digital contents:

  • identifying the acquisition device that produced the data,

  • validating the integrity of the contents,

  • retrieving information from multimedia signals.

The term “forensic” comes directly from the legal environment, where today the traditional understanding of evidence is changing and most prosecutors, lawyers and judges deal with digital evidences, that are intended to include all forms of evidence created, manipulated, stored, transmitted by means of digital devices (such as computers, telephone systems, wireless communication systems, networks as internet, mobile telephones, smart cards, navigation systems, and many others). Whereas the so called Computer Forensics considers the use of scientific methods to extract such digital data by devices involved in criminal scenarios, the Multimedia Forensics, as a second phase, applies scientific methods for the analysis of the contents. The usual methodology is based on the idea that inherent traces (like digital fingerprints) remain in digital content, both during the creation process and any other successive processing; hence, by extracting some digital fingerprints from the data and analyzing their properties, it is possible to have some knowledge on the life cycle of the data.

More specifically, for source identification, forensic algorithms assume that the acquisition device leaves specific traces due to its intrinsic characteristics (for example sensor noise, lens distortion, etc). Basing on such intrinsic characteristics, statistical tests are able to distinguish between computer-generated images, images produced by a scanner and those obtained by a digital camera; distinguish among certain camera models; distinguish which specific camera was used for taking a picture.

Similarly, tamper detection algorithms try to verify the integrity of the content, by assuming that different processing algorithms leave identifiable traces (for example the JPEG blocking artefacts); or that the traces introduced by the acquisition device are altered due to tampering; or that some inconsistencies of scene characteristics are introduced by tampering (for example inconsistencies in light condition).

Finally, Multimedia Forensics considers to exploit a number of signal enhancement procedures (as noise suppression or distortion compensation) in order to attain a higher degree of intelligibility of the data. Furthermore, techniques for information retrieval are applied to multimedia signals, in order to acquire information from visual and audio files, including object colour analysis, pattern recognition, photogrammetric measurement of objects within the scene (for example footprints), anthropometric feature evaluation of people within the scene (for example height), speaker recognition.

 

Research group:

Research area: