一种改进的彩色图像多水印嵌入算法  

An improved multiple watermarking algorithm for color image

作　　者：林椹尠 李源盛 

机构地区：[1]西安邮电大学理学院,陕西西安710121 [2]西安邮电大学通信与信息工程学院,陕西西安710121

出　　处：《西安邮电大学学报》2014年第6期53-57,共5页Journal of Xi’an University of Posts and Telecommunications

摘　　要：针对彩色图像的版权认证问题,提出一种改进的多水印嵌入算法,采用量化索引调制(QIM)量化方法与提升小波变换结合,将水印分别嵌入到高频和低频系数矩阵,即将载体图像转换到YCbCr空间后,对三个分量分别进行提升小波变换,再用QIM量化方法在Cb和Cr分量的高频系数,以及Y分量的低频系数嵌入水印。实验结果表明,利用改进算法嵌入的水印具有较好的隐蔽和不可见性。抗攻击性测试结果显示,在高斯噪声、椒盐噪声、JPEG压缩和1/4裁剪四种攻击方式下,改进算法比基于DCT的纹理特征嵌入算法的水印NC平均值分别提升6%、2%、46%、2%,具有更强的鲁棒性。An improved multiple watermark embedding algorithm is put forward in this paper for the problem of color image copyright authentication.Combined with QIM quantitative and lifting wavelet transform,the watermark are embedded into high frequency and low frequency coefficient matrix.The carrier image is transformed into YCbCr space,the three component into the lifting wavelet transform.Then the QIM quantization method is used in high frequency coefficients of Cb and Cr component,and Y component of low frequency coefficients to embed watermark.Experimental results show that embedded watermark obtained by the improved algorithm has good invisibility and concealment.Anti attack test results show that,compared with the texture feature embedding algorithm based on DCT,the watermark NC average values obtained by the improved algorithm of are increased by 6%,2%,46%,2% under the four attacks of the Gauss noise,salt and pepper noise,JPEG compression and 1/4clipping respectively.Therefore the improved algorithm is more robust.

关 键 词：提升小波变换 多水印嵌入 QIM量化 低频系数 高频系数 

分 类 号：TP309[自动化与计算机技术—计算机系统结构]