Skip to Main Content
The recent increasing demand for 3D digital contents has motivated us to develop techniques for watermarking 3D models so that we can hide copyright information behind them. However, existing methods rather focused on the robustness against possible attacks to destroy the embedded watermarks, and have not fully taken care of the associated influence on the visual quality of the target 3D models. In this paper, we present a new approach to watermarking 3D models based on the visual saliency associated with their shapes. Our algorithm begins by segmenting the perceptually conspicuous regions over the given 3D triangular mesh using surface curvatures, and then selectively embeds the watermark to the regions by statistically modulating the distance between each mesh vertex and the bary center of the mesh. The proposed scheme allows us to hide the watermarks behind the 3D mesh while maximally preserving its visual appearance, and the associated embedded information can be recovered from the watermarked mesh without referring to the original 3D mesh.