A mesh watermarking method based on local roughness analysis

This paper proposes a watermarking algorithm for 3D meshes based on local roughness analysis. The proposed method is a robust and partition-based watermarking method. It is composed of two stages: watermark embedding and extraction. During the watermark embedding stage, the input mesh is decomposed into groups corresponding to different values of roughness, which is measured by the Laplacian of Gaussian curvature. Then, the watermark is embedded in each group individually with a different intensity. In the extraction process, we extract the individual watermark from each group, and then determine the final watermark by using majority vote. According to the experimental results, the proposed method has satisfied both the perceptual-quality-oriented protocol and the geometric-quality-oriented protocol in 3D Mesh Watermarking Benchmark. In addition, it has relatively high payload and is robust against a variety of common attacks, and even cropping. Therefore, this method can further expand the application of 3D watermarking technique.