Research On Robust Watermarking Algorithm For 3D Model

With the rapid development of 3D hardware,the popularity of 3D games and the maturity of 3D printing technology,the 3D model is widely spread and applied.However,while people enjoy the realistic visual perception of the 3D model,it is also easier for the illegal elements to destroy the model.Moreover,the design and construction of 3D models require a lot of manpower,material resources and financial resources.Therefore,it is particularly important for us to study the information security of 3D models and how to enhance copyright protection.Watermarking is currently one of the main ways of copyright protection for 3D models,but limited by the data characteristics of the model itself,the development of the technology still faces many difficulties.Therefore,how to effectively use watermark technology to protect the copyright of the three-dimensional model has become the current research focus.In this thesis,the existing 3D model watermarking algorithm is investigated,and the related research results are deeply analyzed.Then a visible watermarking algorithm and two invisible watermarking algorithms are proposed.The main research work of this article includes:(1)In view of the protection demand of 3D models,this paper takes the logo provided by the design unit as the visible watermark information,and proposes a boundary adaptive visible watermarking algorithm based on the segmentation three dimensional model.By using the TTF character library,the input text is converted to the vertex coordinate information needed to embed.The watermark information is adaptively embedded into the 3D model by the projection,pre subdivision and clipping of the 3D model.It is verified that the algorithm can well display the information contained in Chinese characters,English and numeric information in 3D models,and is robust to common attacks such as noise and subdivision.(2)In view of the fact that most existing watermarking algorithms are not resistant to shear attack,a new watermarking algorithm based on model eigenvalues is proposed in this paper.First,highlight the selected feature points in the 3D model;then,based on the prominent feature points on the combination for segmentation of 3D model vertex information,to ensure that the 3D model segmentation does not repeat,and the vertex information can be fully utilized;finally,through the "two adjustment" to ensure the effective positioning to the corresponding feature point in after embedding the watermark,so as to accurately extract the watermark information.It is proved by experiments that this algorithm can resist many common attacks and can resist the shear attack well.(3)Since the existing watermarking algoritlhm only adopts the single model feature,it is difficult to resist many common 3D model attacks.Therefore,this paper proposes research on enhancing watermark algorithm to the salient point.Using vertex norm and vertex protrusion,the double watermark is embedded in the 3D model,and the robustness of the algorithm is improved.Experimental results show that the proposed algorithm is robust to common affine transformation and signal attack based on good transparency,and it can also resist the simplified attack very well.