Research Of Self-adaptive Blind 3D Model Watermarking

As the development of computer graphics and network technology,the application of 3D model on the network becomes reality,which also brought the issue of copyright protection. Digital watermarking technology as an effective means of copyright protection has recently become a research focus,but most previous research has focused on general types of multimedia data including text data,audio and video.Ohbuchi R is one of the early workers focused on 3D watermarking who proposed several watermarking schemes in 1997.According to whether the original model is needed at detection stage,3D watermark can be divided into blind watermarking and non-blind one.Although blind watermarking is less robustness than non-blind ones,they are more useful for various applications where a host signal is not available.Existing research shows that watermarking intensity gives a better robustness,but normally degrades the visual quality of the watermarked mesh and risks to make the watermark perceptible and vice versa.To achieve a good balance between imperceptibility and robustness,we propose two methods to calculate the roughness under the inspiration of the Cho J.W's work at first:and then we discuss how to design a simple and effective blind watermarking algorithm for 3D model:it maintains the same robustness constraint while the embedding strength adaptive to surface roughness:At last a self-adaptive blind 3D model-watermarking algorithm based on visual masking is proposed.The proposed algorithm uses out-bounding box center instead of gravity to convert 3D model to spherical coordinate,and r coordinate values are divided into distinct sections with equal range.Watermark is embedding into the 3D meshes model by modifying the vertices number ratio of each section.In order to modify the number ratio of the adjacent sections,a set of vertices are selected and ordered according to a minimal distortion visibility threshold at first,and then the selected vertices are modified to embed watermark.The threshold has something to do with the roughness of the model surface and the visual masking effects.Experimental results show that the proposed algorithm has the same robustness as Cho J.W's work,but with lower distortion.