With the introduction of multimedia, documents as pictures, sound recordings, books, etc. are becoming electronic. Copyrighting electronic media is a pressing matter, since digital data can easily be copied. It can be achieved by introducing a (digital) watermark to the original document. Several watermarking schemes have unfortunately been broken. A fundamental question is how to achieve truly secure watermarks. One should take into account that the watermark should not degrade the quality of the original. However, if the watermark can be removed, one can make illegal copies. To solve this problem, the watermark is hidden. A hidden document may be abused by spies for covert communication, which one should prevent as much as possible. Watermarks can also be used to enhance confidentiality of such documents as commercial digital blue prints. In such applications law enforcement wants to trace a receiver of the document who leaked or fold the document to the competition.
This research focuses primarily on: 1. whether watermarks can be designed using only low-bandwidth covert channels, 2. how to generate from the original, modified copies that are computationally indistinguishable from the original one (current research on indistinguishability has mainly focused on producing data which is indistinguishable from a uniform random source), 3. how to make watermarks that can be verified without the use of a computer (a jury nowadays must trust an insecure computer), 4. how to improve tracing algorithms that allow to identify (some of) those involved in the illegal distribution (or leakage) of a document.
