摘要
本文提出了一种基于小波域低频系数量化的彩色图像鲁棒盲水印新算法.首先将RGB彩色宿主图像转换为YCbCr颜色空间,然后对亮度分量Y进行三层小波分解,采用适当的量化间隔对低频逼近子带LL3进行均值量化,根据量化系数的奇偶特性再与二值图像置乱转化后的水印序列进行异或运算,从而把水印嵌入到亮度分量的逼近子带中.仿真实验结果表明,本文提出的算法对JPEG/JPEG2000压缩、添加噪声、滤波、剪切、像素移位等多种攻击有较强的鲁棒性,透明性好,并且能实现水印的盲提取.
The paper proposes a novel robust blind watermarking algorithm for color image based on the DWT low frequency quantized coefficients.The algorithm firstly transforms the color image from RGB color space to YCbCr color space,then to the brightness component Y performs 3-levels wavelet decomposition,and applies mean quantization for the low frequency approximate sub-band LL3 by using a proper quantization interval,finally casts the binary watermark sequences into the approximate sub-band of the component Y through the XOR algorithm between the parity properties of quantized coefficients and the watermark sequences scrambled.The experiment results show that the algorithm proposed in this paper is very robust to the common attacks such as JPEG/JPEG2000 compression,additive noise,filtering,cropping,pixel shifting etc,and has a good transparency.What's more,it can achieve the detection of watermark without the original image.
出处
《电脑知识与技术（过刊）》
2007年第14期540-541,580,共3页
Computer Knowledge and Technology
