A suitable watermarking scheme has to be sufficiently reliable such that it is feasible to use it as a mean to resolve disputes over ownership or origin of a particular object in the context of a court of law or similar situations. It should be infeasible to remove the watermark from the object on purpose by a potential attacker or unintentionally by a common transformation of the object. Ideally any transformation of a watermarked object which still contains the original, valuable content should also contain the watermark. If a transformation removes the watermark the resulting object should be useless with respect to its original intended use. Furthermore, the watermark should also be invisible in the sense that it does not interfere with the object’s purpose.

Note that we implicitly assume that the object contains valuable information which is suitable for watermark protection. It is, however, important that a reasonably clear notion of what this valuable information is and in what context it should be protected by a watermark exists. Requiring to protect any transformation of the given object is not sufficient and not achievable as arbitrary transformations can result into arbitrary objects. In general the set of valuable transformations of an object can be regarded as a neighbourhood of this object, but this neighbourhood has to be defined more clearly in order to protect it effectively. The neighbourhood is given by a set of transformations that the object may undergo without removing the valuable information.

Cite this page as 'Frank C Langbein, "Watermarking," Ex Tenebris Scientia, 21st December 2004, https://langbein.org/watermarking/ [accessed 29th September 2022]'.

CC BY-NC-SA 4.0 This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.