The Research Of Image Digital Blind Watermarks Algorithm Based On Wavelet Transform

Along with the Internet popularization and the development of multimedia techniques, because the security of multimedia information and its right protection is currently a popular research topic, digital watermark technique emerges as the times require. Wavelet transform is becoming mature technology, and it has transform mechanism coinciding to the visual characteristics of human visual system. The new image compression standard JPEG2000 and new video compression standard MPEG-4 are taking the system of wavelet transform. This paper focuses on the research of the robust blind watermark algorithm of wavelet transform. The main tasks are:(1) Firstly, digital watermark and related theories and technologies are introduced in detail, and the robustness of typical digital image watermark algorithms is analyzed. Secondly, 3-level wavelet decomposition and reconstruction are performed for images in the Mallet algorithm. Furthermore, we analyze different levels of wavelet coefficient, energy distributions and robustness presented out from the similarity of 3-level wavelet coefficient blocks by experiment. Finally, the embedded position of watermark is analyzed. On the basis of above all, research thought is given and the key problem, which needs to be solved in this paper, is then induced.(2) A robust blind watermark algorithm based on the similarity of wavelet coefficient blocks is proposed. Firstly, host image is decomposed using 3-level wavelet transform, and its low frequency coefficient sub-band is dispersed in 2×2 blocks. The ring model of most similar blocks is established by making use of Euclidean distance as comparability measurement. The embedded position of watermark is chosen according to the size relationship between information entropies of two neighboring blocks, and low frequency coefficient are modified to control embedded watermark strength. Finally, the invisibility and robustness of our proposed algorithm are analyzed through experiments.(3) The self-adaptive robust blind watermark algorithm based on wavelet coefficient tree is proposed. Firstly, decompose the original image using 3-level wavelet transform, and construct a wavelet zerotree tree with its low frequency coefficient as root. Secondly, choose two trees, whose zerotree roots are of comparatively different JND value. Sequentially modifying the coefficient of both trees, the size relationship between weight values of both trees are changed to self-adaptively embed watermark.. Finally, the invisibility and robustness of our proposed algorithm are analyzed through experiments.