›› 2011, Vol. 24 ›› Issue (5): 112-.

• 论文 • 上一篇    下一篇

基于DCT与SVD相结合的图像水印新算法

吕东岳,徐贵力,李开宇,田裕鹏   

  1. (南京航空航天大学 自动化学院,江苏 南京 210016)
  • 出版日期:2011-05-15 发布日期:2011-05-19
  • 作者简介:吕东岳(1986-),男,硕士研究生。研究方向:图像隐藏与数字图像水印技术。徐贵力(1972-),男,教授,硕士生导师。研究方向:领域机器视觉,图像处理。李开宇(1969-),男,副教授,硕士生导师。研究方向:计算机测控技术,数字图像处理。田裕鹏(1964-),男,教授,硕士生导师。研究方向:数字图象处理及其应用。

A New Image Watermarking Algorithm Based on DCT and SVD

 LV Dong-Yue, XU Gui-Li, LI Kai-Yu, TIAN Yu-Peng   

  1. (College of Automation,Nanjing University of Aeronautics and Astronautics,Nanjing 210016,China)
  • Online:2011-05-15 Published:2011-05-19

摘要:

基于奇异值分解的水印算法中,利用正交矩阵实现水印的嵌入保证了水印的不可替换性。但在修改正交矩阵时,如果不保持矩阵的正交性,就会对提取出水印的正确率产生影响。为了解决这一问题,提出了一种基于二维离散余弦变换与奇异值分解相结合的数字图像水印新算法。在水印嵌入时,对图像子块的DCT系数矩阵进行SVD分解,利用三角函数的正交性修改得到的正交矩阵实现水印的嵌入。实验表明,与现有利用奇异值正交矩阵的方法相比,由于在嵌入水印时保持了矩阵的正交性,从而使其抵抗JPEG与椒盐噪声攻击能力更强,提取水印的正确率更高。

关键词: 离散余弦变换, 奇异值分解, 正交矩阵, 水印

Abstract:

Among all watermarking algorithms based on singular value decomposition (SVD),using the orthogonal matrices assures non-replacement of watermark,but it will affect the accuracy of the watermark if the orthogonality is not maintained.In order to solve this problem,a novel binary watermark algorithm based on discrete cosine transform (DCT) and SVD is proposed.The left orthogonal matrices of host image's sub-blocks' DCT coefficients are modified according to the orthogonality of trigonometric function to embed watermark bits in the proposed algorithm.Experimental results show that compared with current algorithms using SVD,the orthogonality of the left orthogonal matrices is maintained.The results also show higher extracting accuracy in resisting JPEG compression and salt and pepper noise.

Key words: discrete cosine transform;singular value decomposition;orthogonal matrix;watermark

中图分类号: 

  • TP391