基于离散余弦变换与Hessenberg分解的图像水印算法

为了增强水印系统的视觉隐秘性与抗几何攻击能力,设计了基于离散小波变换与Hessenberg分解的图像水印算法。将宿主图像实施分割,形成一系列的8×8的非重叠子块;随后,联合信息熵与边缘熵值,构建嵌入区域选择方法,从这些子块中确定出合适嵌入水印的子块;引入离散余弦变换DCT(Discrete Cosine Transform),对选择的子块进行分解,输出对应的低频与高频系数,将每个子块的低频系数组合成相应的矩阵;随后,利用Hessenberg分解方法处理这些低频系数矩阵,输出对应的上Hessenberg矩阵;根据结构相似度与位正确率,构建粒子群算法的适应度函数,通过对其迭代,获取最优的嵌入强度;根据优化的嵌入强度,设计水印嵌入机制,将水印信息隐藏到上Hessenberg矩阵中,形成水印图像;最后,构建水印检测方法,从水印图像中提取水印。实验数据表明:较当前鲁棒水印方案而言,所提技术具有理想的视觉不可感知性与鲁棒性。

IMAGE WATERMARKING ALGORITHM BASED ON DISCRETE COSINE TRANSFORM AND HESSENBERG DECOMPOSITION

In order to enhance the visual concealment and anti-geometric attack ability of the watermarking system, an image watermarking algorithm based on discrete wavelet transform and Hessenberg decomposition is proposed in this paper. Firstly, the host image is segmented to form a series of 8 * 8 non-overlapping blocks. Subsequently, by combining information entropy and edge entropy, the embedding region selection method is constructed to determine the appropriate embedding watermarking sub-blocks from these sub-blocks. Discrete cosine transform is introduced to decompose the selected sub-blocks for obtaining the corresponding low-frequency and high-frequency coefficients, and the low-frequency coefficients of each sub-block are combined into corresponding matrices. Then these low-frequency coefficient matrices are processed by using Hessenberg decomposition method to output the corresponding upper Hessenberg matrix. According to structural similarity and location accuracy, the fitness function of particle swarm optimization is constructed to obtain the optimal embedding strength through iteration. The watermarking embedding mechanism is designed base on the optimized embedding strength to hide the watermarking information into the upper Hessenberg matrix for obtaining the watermarking image. Finally, the watermark detection method is constructed to restore the watermark from the watermark image. The experimental data show that the proposed technique has ideal visual imperceptibility and robustness compared with the current robust watermarking schemes.