Research On Zero Point Cloud Model Watermarking Algorithm

Recent years, with the development of digital technology and network technology, various forms of digital products via the Internet have spread with fast propagation speed and wide range. As a result, it becomes very easy to copy, modify, and plagiarize between products. Such a low-cost, high-speed mode of transmission often leads to various degrees of infringement. Therefore, in order to protect the intellectual property rights of digital products, digital watermarking technology came into being. Digital watermarking technology is classified in accordance with the actual embedded watermark information you need. It is can be divided into the embedded watermark and zero watermark technology. The embedded watermark technology uses the insensitivity of the human sense organs and the redundancy of digital product itself by adding a slight disturbance in the airspace or transform domain to embed the watermark, in most cases watermark only can be embedded in low intensity, thereby reducing the robustness of the watermark. While the zero-watermark breakthrough conventional embedded watermark algorithm in approach to modify the model data, directly use the airspace of the model itself or transform domain characteristics to construct the watermark information to solve the contradiction between watermark robustness and transparency. On the basis of spectral analysis to achieve copyright protection model can well solve the contradiction between watermark robustness and transparency of watermarking.Starting from the point of view of copyright protection of3D point cloud model, I proposed a robust zero-watermarking algorithm based on point cloud model. Research as follows:1. In order to be able to efficient processing point cloud model data, reduce the difficulty of the post-data processing, we need to create a model octree data structure to split model by the spatial structure.2. Point cloud model just discrete set of points in space. They only have point coordinate information not the topology connecting among the points. By setting up Kd-tree data structure of each block in model, we can know the relationship between topological connections.3. In order to obtain the important features of the model itself, respectively the spatial information within the model of each sub-block transformed to the frequency domain to draw the spectral coefficients.4. Low-frequency spectral coefficients concentrated most of the energy of the model, so we take up the average structure of the low-frequency spectral coefficients in each sub-block to form1bit watermark information.5. Conducting a variety of attacks to test the robustness of the watermark algorithm. Experiments show that the algorithm to affine transform attacks, noise attack and simplify attack is strongly robust.