A spatial and DCT-domain composite watermarking of three-dimensional triangle meshes

We propose a new scheme for digital watermarking of three-dimensional (3-D) triangle meshes. In order to insert and extract watermark signals, we generate triangle strips by traversing the 3-D mesh model. We split the strips into two groups. For the first groups we embed the watermark signal into vertex positions of the 3-D model in the DCT domain, and for the second group, we embed the watermark in the vertex positions in the spherical coordinates. In both cases, we apply masking operation for the watermark signal according to the variation of the host data along the traversed strips so that we can change the embedding strength adaptively, and reduce the visibility distortion of the data embedding. We test robustness of the watermarking scheme by embedding a random binary sequence and applying different attacks, such as additive random noise, compression by MPEG-4 SNHC, and affine transform.