Research On Digital Watermarking Algorithm Of Still Image

As the rapid development of digital communications network and the extensive application of digital multimedia, information security and copyright protection of digital works address an urgent need to the practical problems. Digital watermarking technology has been a solution in the digital multimedia content copyright issues and the principal instrument of certification, and it is the focus of academic research. This paper first introduces the digital watermark technology background, classification, application and research status, gives generic model of the watermarking, developments digital watermarking technology algorithm design requirements and their mutual relations and also introduces the assessment watermarking system and different watermark algorithm and analysis of the classification. Then orients the still images , with the image feature recognition, image segmentation, image scrambling, DFT transform, and other technology and methods, two digital image watermarking scheme are given. The first scheme is based on the transform domain, the main features are as follows: (1) use of space domain and frequency domain methods: in the space domain information from the characteristics which represent the main image, and in accordance with the density of characteristics to do an image segmentation, to meet the requirements of the image block watermarking embedded in the frequency domain. As well as with the transparence in space domain, frequency domain reached the robustness. (2) The algorithm is based on image content: embedding the feature point - intensive regional, and sub-image features bundled together, so they can resist the common signal processing attack. (3) By the feature extraction of the characteristics of the surrounding neighborhood and feature vectors similarity matching, we restore the geometric transformation watermark image, and obtain certain geometric attack robustness.The other watermarking algorithm is based on adaptive feature recognition of the spatial domain. The algorithm uses a pre-attack method, and adaptively adjusts the threshold of Harris corner detector until a certain number of feature points are got. The pre-attack process considers all the existing major watermark attacks, so excellent stability of the feature points is reached and the points can be used to estimate the parameters of geometric attacks and restore the watermark sync, combining the inherent advantages of feature points, the algorithm has good robustness, especially for geometric attacks. This algorithm made full use of regional characteristics of image, and the contradiction between robustness and perception can not be solved effectively.In this paper, we have made a large number of simulation experiments with matlab. We use StirMark to simulate various attacks to watermarked image. The simulation results show that both the algorithms have good robustness.Finally, this paper gives a detailed summary of the work, and puts forward the future work needed to be improved.