摘要
提出了一种结合奇异值分解(SVD)和DCT变换的零水印算法。其核心思想是合理选取阈值,根据该阈值对图像块进行相应的奇异值分解或DCT变换,然后利用几个重要系数构造原始图像的提取水印信息,同时又从外部加载水印信息,既满足图像认证的需要,又能用来进行数据隐藏。同时针对现有水印技术大都难以抵抗几何攻击的现象,利用一个或多个几何矩对经过几何变换的水印图像进行参数估计并校正。实验表明此算法在压缩、滤波、几何变换等方面都具有很强的鲁棒性。
The paper brings forward a kind of hybrid SVD-DCT zero-watermark algorithm. The central idea of this method is: the threshold value is selected correctly firstly to decide which transform should be used on each block, then extracted watermark is formed by several important coefficients and adds external watermark. Not only can this meet the needs of the image authentication, but also can be used to hide the data. As many proposed image watermarking techniques are sensitive to geometric distortions, the geometric moments of original image are exploited to estimate the geometric distortions. So that the corrupted watermarked image can be corrected and the watermark can be detected correctly. The obtained results show that the technique is robust to most attacks including JPEG compression, filter and geometric transform.
出处
《金陵科技学院学报》
2004年第4期3-6,共4页
Journal of Jinling Institute of Technology
