Research Of Image Digital Watermarking Algorithm Based On Object

The digitalization of multimedia data has greatly provided the convenience of the storage of multimedia information, and a variety of data can be uploaded freely to the Internet. However, the private problem has become more and more serious. Digital watermarking technology is an effective way to solve the problem by hiding secret information in the digital multimedia (image, audio, video etc). Moreover, watermark algorithms based on transformation domain have the capacity to resist the signal processing and the harmful attacks, and some algorithms have also integrated with the compress standard. The research of watermark algorithms based on transform domain have a good prospect of application.The thesis makes the systematical researches on the digital watermarking algorithm based on resisting geometric attacks. First, the background and the current situation of digital watermarking are introduced, and the basic concept and character, the universal models, the typical algorithms, the attack analysis and performance evaluation of digital watermarking techniques are also introduced. Then the thesis lays great emphasis on the technology of the image watermark to resist geometric attacks, some classical image watermark algorithms are given, and their individual application environments, the advantages and the disadvantages are analyzed.Enlightened by the idea of the second generation watermarking technology, a new algorithm is proposed with image digital watermarking resisting geometric attacks based on object. For implementing the algorithm, we introduce the principle of the second generation watermarking technology at first, then propose the algorithmic which based on the idea of embedding watermark in a certain key object of the image, and present steps such as edge detection, object extraction and determining. The object extraction is carried out by three standards for distinguishing the background and object. After Arnold transformation, the watermarking will be embedded into the DCT domain of the original image. While the watermarking is extracted, the log-polar coordinate transformation is introduced into the algorithm, which improve the robustness of resisting geometric attacks. Simulation results show that this algorithm can be immune to rotation, scaling, low pass filtering and cutting. Finally, the entire work is summarized and the direction of the future research on watermarking is given.