December 5th
2pm, room Byron beige
By Xavier Rolland-Nevière
The need for robust and blind watermarking techniques for triangular meshes has grown alongside their ever more pervasive use in, e.g., video games and movie productions. One of the most successful approaches consists in altering the distances between the vertices and the center of mass of the mesh. Such an approach has recently been formulated as a quadratic programming problem, which has shown to yield significant benefits. In this talk I will present a generalization of this formulation, with both integral quantities and arbitrary relocation directions for the vertices, while still using the effective quadratic programming framework. The results of a comprehensive benchmark of our generalized 3D watermarking procedure against mesh alteration procedures show a significant leverage with regard to e.g. the simplification attack.
