The Research About Adaptive Watermarking Algorithm For Images Based On Wavelet Domain

With the rapid development of computer, network and communication, especially the popularization of the Internet, information security and protection have been a key and attractive problem. So, digital median work needs an effective method of the copyright protection, and in order to that, digital watermarking has attracted a lot of attention. In wavelet domain, adaptive digital watermarking technology for image protection is researched in the dissertation. At first the notation, principle, generic model, basic characteristic, typical algorithms, attack analyses and evaluation of performance are described. Then the theory and application of wavelet transform are introduced. Based on the research and the problems that exist to be considered and solved, four adaptive watermarking algorithms in wavelet domain for images are proposed. They are all about the problem of how to get the best embedding strength of the watermark and embed the watermark adaptively. The four are list in detail as follow.Firstly, considering the features of human visual system (HVS) and the wavelet transform, a new adaptive watermarking algorithm is proposed. The original image is decomposed by wavelet transformation. Then wavelet coefficients are organized in order to form wavelet blocks. Based on several parameters with the features of human visual system (HVS), they are classified into several classes. Watermark components with different strength are embedded into different wavelet blocks based on the results of classification and the features of wavelet transform.Secondly, based on wavelet packet transform and fuzzy blocks classification, a novel adaptive watermarking algorithm is proposed. The original image is converted to frequency domain with wavelet packet transform by using the m-sequence. Then some wavelet coefficients are organized to form wavelet blocks. After that, these blocks are classified by the model based on human visual system(HVS) and the model based on energies. Finally binary watermark components with different strength are embedded to different wavelet blocks according to the results of classification.Thirdly, considering neural-fuzzy control, a novel adaptive watermarking algorithm is proposed for image. The original image is decomposed by wavelet transformation. And then some wavelet coefficients are organized to form wavelet blocks. After that, the neural- fuzzy control network is set up and trained in order to get the best embedding strength of the watermark. Finally, watermark components with different strength are embedded to different wavelet blocks according to the best strength got from the trained network.Lastly a novel adaptive blind watermarking algorithm is proposed. The data payload of gray-level watermark can be greatly reduced by DWT compression. Moreover, the selection of the embedding place and the strength of watermark are self-adaptive to the image. The watermark can be extracted without the original image and the parameters used in embedding process.All experiment data is drawn out for the performance analyses. The experimental results show that the imperceptibility and robustness of these methods can be guaranteed. Finally we review entire work. Something need to research in the next step is pointed out. And we suggest the direction for the future research of watermarking.