【摘要】 提出了一种规范类正交矩阵,该矩阵中任意2行的互相关系数为定值,且矩阵列数不受限制.首先,将隐秘数据经过规范类正交矩阵编码,嵌入载体图像的2×2分块离散余弦变换系数矩阵的同频分量内,以提高信息隐藏算法的安全性.其次,基于参量数据的贪心算法,实现规范类正交矩阵的快速构造.最后,利用邻域信息的相似性,实现隐秘数据的正确提取.实验数据的分析结果表明,信息嵌入量、组块数、互相关参数、检测错误率等指标之间存在相互制约的关系.模拟结果表明,所提出的信息隐藏算法具有图像失真小、嵌入容量大、鲁棒性强的特点.更多还原

【Abstract】 A kind of normal similar-orthogonal(NS) matrices is defined.The cross-correlation coefficient of any two rows in the NS matrices is a constant and the number of columns has no bounds.First,the secret information is coded by a NS matrix and embedded into the same frequency components of 2×2 discrete cosine transform coefficients of the original image,which can improve the security of the information hiding algorithm.Then,by introducing a greedy algorithm based on parameters,the fast generator of NS matrices is realized.Finally,by using image neighboring relativity,the secret information can be totally extracted.The analysis results of empirical data show that there are trade-off relations among the embedded information number,the group number,the cross-correlation parameter and the detection error probability.The simulation results demonstrate that the proposed method has low image distortion,high embedding capacity and high robust.更多还原