基于DCT和SVD变换的盲数字水印算法

Blind digital watermarking algorithm based on DCT and SVD transform

结合奇异值分解(SVD)变换,提出一种基于DCT域内DC分量的数字盲水印算法。对原图像进行8×8分块DCT变换,对DC系数集合进行2×2分块的SVD变换,使用量化步长Q量化每个小块对应的最大奇异值,向量化后的结果嵌入水印。由于选择DC系数进行SVD变换,且SVD变换结果表现的是图像的内蕴特性而非视觉特性,故算法具有较高的稳健性和透明性。实验表明,该算法可以有效地抵抗中值滤波、JPEG压缩、噪声等常规篡改处理,水印的提取不需要原始图像参与,执行起来简单方便。

Combining SVD transform,this paper proposes a blind digital watermarking algorithm based on DC coefficients of DCT-domain.This algorithm executes DCT transform on each 8×8 block of the original image,and executes SVD transform on each 2×2 block of the collection of the DC coefficients.The quantization step Q is used to quantify the max singular value in order to embed the watermark in the result of the quantization.Because the DC coefficients are chosen for SVD transform,and the results of SVD represent the internal properties instead of the visual features,this algorithm gets a better robustness and transparency.The experiment results show that it can effectively resist the tampering attacks such as：median filtering,JPEG compression,noise and so on.The extraction process becomes very easy and convenient because it does not require the original image to involve in.