Improved Blind Watermarking of 3D Objects Based on Partition

With the development of architecture design, virtual reality, entertainment, mechanical engineering and cultural heritage, digital watermarking has played an important role in copyright protection of 3-D objects. However, due to the problems such as computational complexity and capturing local geometrical feature of 3D mesh, watermarking of 3D objects has not reached the same level of maturity as that of still images and video. In order to alleviate them, mesh partition is applied in the algorithm. This paper presents a novel mesh partition approach which can predict the partition boundary of stego-mesh with some additional information instead of the presence of the cover-mesh. The canonical positions of meshes are determined by employing the face normal principal component analysis. So the additional information is picked up scatteredly from the elevation of the cover-mesh in canonical coordinate systems and used to match the elevation of the stego-mesh. After matching the elevation, the deflected degree of coordinate system is estimated to realign the stego-mesh approximately and plot out the partition boundary. Finally, an improved watermarking is proposed and a group of experiment results are presented to verify the stability of the proposed partition scheme against some attacks imposed on meshes.