Research On Robust 3D Watermarking Algorithm

In 1993,Tirkel proposed the concept of "digital watermarking".Digital watermarking is fully used in the hidden logo,anti-counterfeiting traceability,authentication and security covert communications and other fields.In recent years,with the rapid increase of speed of computer processing and the advantages of the realistic image of the 3D model,3D model has been spread widely.As a variety of important copyright protection technology,3D model watermarking is still in the initial stage.Therefore,three-dimensional model blind watermarking algorithm with strong robustness,high transparency,and safe blind extraction mode is studied,which has become an important direction for the development of 3D watermarking.Firstly,this thesis describes the background of digital watermarking technology,basic concepts,and the main purpose.Followed by the analysis of 3D watermarking development present situation,this thesis summarizes the present mode of attack and introduce evaluation methods in performance from the aspects of transparency,robustness,embedding capacity.Then this thesis proposes a combining airspace and transform domain of 3D watermarking algorithm.In recent years,many of the airspace and transform domain algorithms are proposed,some of the watermarking algorithms have strong robustness to some watermark attacks,but performan generally to other attacks.As a result,proposed algorithm effectively combines the airspace and transform domain watermarking algorithm,in order to effectively resist various watermark attack with enhance advantages and weaken the disadvantage.In the specific algorithm design,we choose the target vertex according to the size of the curvature.this thesis use the watermarking algorithm based on the aspect ratio at the adjoining point of the vertex,use the watermarking algorithm based on the Laplace matrix at the bicyclic and tricyclic neighbors,The watermarking algorithm based on area ratio is used on the ring adjacency.Three watermarking algorithms embed the same watermark information,in order to improve the robustness.Finally,this thesis proposes a 3D watermarking algorithm based on spherical mapping.The main idea of the algorithm is to map the 3D model onto the sphere,and then map to the gray scale image according to the spherical coordinates.Then,the pixel of the gray image is replaced,and the watermark information is embedded by changing the gray value of each pixel.The three-dimensional model is mapped to the two-dimensional model for processing,not only can the existing image watermarking algorithm,but also the calculation is small,but the model mapping location is susceptible to interference,when the noise attack,the model mapping location may appear offset.In order to overcome this shortcoming,this thesis uses K-means to segment the three-dimensional model,which can be divided into K parts according to the actual needs.The K parts are relatively stable,which can reduce the influence of attack on the mapping position.At the time of extraction,the K-means method is used to classify the stored points as the initial center.In order to solve this problem,we can save the segmentation fragment and the watermark In addition,according to the spherical coordinates of the gray image,you can further reduce the impact of the peak position of the model.