基于能量估计和视觉模型的数字水印算法

本文将数字水印能量估计方法拓展到分块离散余弦变换 (DCT)域,并且提出了一种基于分块DCT的数字水印算法,将能量估计方法应用于该算法中。该算法利用了人眼视觉系统 (HVS)模型,将水印信息嵌入到DCT系数的低、中频域,试图达到不可感知性和鲁棒性的最佳折衷,进而提出了一种灵活的重复嵌入策略及相应的检测算法。实验结果表明,该算法在较大的信息嵌入量情况下仍然具有较好的不可感知性和对多种攻击的鲁棒性。

A Watermarking Algorithm Based on Energy Estimation and Human Visual System Model

In this paper, the watermarking framework for energy estimation is extended to block-based Discrete Cosin Tranform (DCT) field. A watermarking algorithm based on the block-based DCT is presented, and the watermarking energy can be determined from the given Power Signal-to-Noise Ratio(PSNR). The algorithm, inserting the watermarking into the low and then middle frequency field of DCT coefficient based on Human Visual System (HVS) model, tries to get a good trade-off between imperceptibility and robustness. It also gives a re-embedding and correspond detecting method, which makes it can resist against various attacks. The result of experiment shows that the algorithm does get a good trade-off between imperceptibility and robustness.