Research On Digital Image Watermarking Algorithm Based On Wavelet Transform

With the development of the computer network and the communication, the digital media, including digital image, digital video,and digital audio, has been applied widely. However, security of digital media information, rights protection of intellectual property and authentication problems, which have changed prominently, becomes an extremely urgent important subject in the digital world. The digital watermark, which has become an effective supplement method of the tradition encryption, is a new technology that can protect the copyright and the authentication origin in the open network environment. Especially, in recent years it had drew more and more attention in the society, and has become a research hot spot in the field of multimedia signal processing.This thesis reviews some fundamental theories of digital watermarking, for which carried out in-depth discussion on the research of digital image watermarking technology. Based on image scrambling and wavelet transform, algorithm is systematically studied. Moreover, image scrambling technology is used for pretreatment of watermark information, and arnold image scrambling algorithm is improved. On the one hand, the combining the characteristics of human visual masking of embedding the watermark image, it has better adjusted contradiction between the watermark invisibility and robustness. On the other hand, the results of experiment show this algorithm has a high peak signal noise ratio, and has increased the invisibility of the watermark. For compression, noise, and other attacks, this algorithm has better robustness.In view of the above algorithm, extracting watermark still need the original image. Based on discrete wavelet transform, blind watermarking algorithm is proposed and designed.Firstly, host image is blocked, in order to reduce effectively the time complexity of algorithm.Secondly, watermark information is embedded in important coefficients of image. Related informations of watermark image are compared with threshold, then watermark information is embedded in high-frequency part of subgraph and low-frequency part of subgraph. Finally, the results of experiment show that it has better robust for common attacks such as shear and JPEG compression. But the disadvantages require further improvement in transparency of images and robustness of noise filtering attacking.