Research On Digital Watermarking Algorithm Based On Wavelet Analysis

On the base of research and discussion of digital watermarking algorithms, the thesis designs and realizes the still image digital watermarking algorithm in spatial and frequency domain, and the robustness is also tested. The fundamentals, such as basic features, principles, classification and the current development status of digital watermarking are analyzed, and the general frame of digital watermarking is given, which provide theoretical basis for the digital watermarking algorithm's proposal, realization and testing. As the base of algorithmic theorem of digital watermarking algorithms, the process of development, basic definition, multi-resolution analysis of wavelet, Mallat algorithm and its application in the image processing are introduced. What's more, two-dimension wavelet decomposition and reconstruction algorithms for the still image are given. As the processing of watermarking-scrambling is taken into account, image-scrambling algorithms such as Arnold transformation, sampling theorem, Hilbert transformation are introduced and simulated. Although spatial algorithm's robustness is not good, it is important in the secret communication. The algorithms based on image emerged and LSB replaced are given and tested. Two different digital watermarking algorithms are given, which are based on DCT domain. One embeds digital watermarking into DCT domain's middle frequency component, the other, associated with pseudo number series and significant watermarking, is based on correlative check in DCT domain. Especially, the latter was improved. The experimental results show that both of them can resist signal distortions such as JPEG compression, Gauss noise, and the improved algorithm has better robustness and security. For present popular wavelet domain algorithms, the watermarking and the embedding algorithms in low and middle domains, associated with image scrambling and singular value decomposition algorithms, are proposed and tested. The results of computer simulation show that the proposed watermarking algorithms satisfy the requirements of robustness and transparency. Both of schemes can resist the common image processing operations, such as lost compression, filtering and noise.