In this paper, we present a new robust database watermarking scheme the originality of which stands on a semantic control of the data distortion and on the extension of quantization index modulation (QIM) to circular histograms of numerical attributes. The semantic distortion control of the embedding process we propose relies on the identification of existing semantic links in between values of attributes in a tuple by means of an ontology. By doing so, we avoid incoherent or very rare record occurrences which may bias data interpretation or betray the presence of the watermark. In a second time, we adapt QIM to database watermarking. Watermark embedding is conducted by modulating the relative angular position of the circular histogram center of mass of one numerical attribute. We theoretically demonstrate the robustness performance of our scheme against most common attacks (i.e., tuple insertion and deletion). This makes it suitable for copyright protection, owner identification, or traitor tracing purposes. We further verify experimentally these theoretical limits within the framework of a medical database of more than one half million of inpatient hospital stay records. Under the assumption imposed by the central limit theorem, experimental results fit the theory. We also compare our approach with two efficient schemes so as to prove its benefits.

In this paper, we adapt the robust reversible watermarking modulation originally proposed by Vleeschouwer for images to the protection of relational databases. The resulting scheme modulates the relative angular position of the circular histogram center of mass of one numerical attribute for message embedding. It can be used for verifying database authentication as well as for traceability when identifying database origin after it has been modified. Beyond the application framework, we theoretically evaluate the performance of our scheme in terms of capacity, distortion, and robustness against two common database modifications: 1) addition and 2) removal of tuples. To that end, we model the impact of the embedding process and of database modifications on the probability distribution of the center of mass position. We further verify experimentally these theoretical limits within the framework of a medical database of more than one million of inpatient hospital stay records. We show that under the assumptions imposed by the central limit theorem, experimental results fit the theory. We also compare our approach with two recent and efficient schemes so as to prove its benefits.

3D video content has been receiving increasing interest over the last few months and created challenges regarding how to protect such high valued items. For instance, depth-image-based rendering techniques allowing for the creation of virtual views may impair underlying watermarks embedded within individual views. In contrast with previous works, this article focus on defining a view-independent watermark embedding domain rather than a watermarking system coherent with the disparity across several views, thus permitting blind detection. More specifically, based on the fact that pixel displacements across views most often reduce to horizontal shifts, the average luminance value along the rows of the video frames have been selected for watermark embedding. Experimental results clearly prove the robustness of the watermark against virtual views generation.