基于分块三维小波变换的盲视频水印算法  

Blind video watermarking algorithm based on 3D wavelet

作　　者：徐达文[1] 王让定[2] 

机构地区：[1]宁波工程学院电信学院,宁波315016 [2]宁波大学纵横智能软件研究所,宁波315211

出　　处：《东南大学学报（自然科学版）》2007年第A01期201-205,共5页Journal of Southeast University：Natural Science Edition

基　　金：国家自然科学基金资助项目(60672070);宁波市自然科学基金资助项目(2006A610018)

摘　　要：针对数字视频的版权保护,提出了一种基于分块三维小波变换的视频水印算法.将视频信号分成三维图像块,根据人类视觉系统的特性,选择纹理复杂的运动块进行三维小波变换.对水印信息进行扩频CDMA编码后,将其嵌入到三维块的小波系数中.利用正交码的自相关函数特性来检测水印信息,实现水印的盲提取.实验结果表明,该水印方案在保证视频视觉质量的同时,对于针对视频水印的特殊攻击具有高鲁棒性.For the copyright protection of digital video, a video watermarking algorithm based on 3D wavelet transform is proposed. The original video frames are splitted into 3D-blocks, the motional block with complex textures used to embed the watermark is selected according to human visual system （HVS） characteristics. Then, the code division multiple accessing （CDMA） encoded watermark is embedded into the wavelet coefficients of the selected sub-block. Blind recovery of the embedded data is achieved using the auto-correlation of the orthogonal codes. Experimental results show that the watermarked frames are indistinguishable from the original frames subjectively and the proposed video watermarking algorithm is robust against the attacks of additive Gaussian noise, frame dropping, frame averaging and lossy compression.

关 键 词：视频水印 三维小波变换 码分多址通信 人类视觉系统 

分 类 号：TP309.2[自动化与计算机技术—计算机系统结构]