Research On Digital Watermark Algorithm Based On 3D Model

With the development of digital media information processing technology and the expansion of applications, the issues of author copyright protection become more and more important. Digital watermarking technique is a good way to solve this problem. Owing to the three-dimensional model (3D) data is increasingly becoming popular multimedia data types, used to Internet the data exchange and visualization, three-dimensional model digital watermarking technology will be used to solve the growing copyright issues.This dissertation first introduces the purpose of the 3D model digital watermarking technology and significance of the protection of the copyright research. Moreover an overview of the current status of 3D model digital watermarking technology development at home and abroad. Then introduces the modeling methods and data representation, the basic framework of 3D model digital watermarking, the basic characteristics, the problems encountered in the process of the research, the performance evaluation of 3D model digital watermarking technology and different attacks. Lastly the classification of 3D model digital watermarking algorithm is introduced.In the context of 3D model digital watermarking technology at home and abroad. The thesis proposes a block vector based improved algorithm of 3D mesh model digital watermarking. The algorithm is to cut the 3D mesh model, replaces center of the whole 3D model using center of each patch, then constructs the basic geometric unit for each patch. At the basis of incurring minimal cost to visual sensation system, it adjusts center of each patch in order to embed bit watermark information ,and watermark spread over the whole surface of the model. This algorithm has well robust to different attacks such as noise, simplify, affine transformations, especially the crop attack. In addition, the paper puts forward a Laplacian matrix based 3D model digital watermarking improved algorithm. According to the topology of vertices of 3D model, the thesis constructs Laplacian matrix, and then carries on a spectral analysis, in the basis of which adjusts the spectrum coefficient which will embed the watermark information into three-dimensional model. Before watermark is embedded, it needs a pre-treatment using spread-spectrum principle, thereby increasing robustness to the random noise attack of the watermark. Lastly the article uses image watermark information instead of simple watermark signal information, experimental results show that the algorithm can resist different attacks, such as affine transformation, the random noise of vertex coordinates, the network smooth (low-pass filtering of the grid) etc, but has poor resistance to crop attack. Therefore, the paper combines two kinds of algorithms, proposes block-based vector Laplacian matrix algorithm, and makes a pre-treatment for grayscale watermark image using replacement algorithms, thereby complements the resistant ability of each other while network attacked, and then improves the robustness of the algorithm.