基于MPEG-7形状轮廓描述编码的数字水印算法  

A NOVEL VIDEO WATERMARKING ALGORITHM BASED ON MPEG-7 CONTOUR DESCRIPTION

作　　者：黄斌[1] 陈德礼[1] 

机构地区：[1]莆田学院电子信息工程系,福建莆田351100

出　　处：《陕西科技大学学报（自然科学版）》2008年第6期119-123,共5页Journal of Shaanxi University of Science & Technology

基　　金：福建省自然科学基金(2008F50602)

摘　　要：为了更好的进行视频数据的版权保护,作者提出了一种基于MPEG-7轮廓描述编码的视频水印算法.在该算法中,原始视频数据被分割成等帧数的视频组,利用MPEG-7的轮廓编码计算每个视频帧的主要物体轮廓坐标点集合;接着通过傅立叶变换得到该帧轮廓点形状不变量,利用哈希的方法计算水印信号并将这些不变量的有效值作为密钥之一;再对整个视频组的轮廓点进行傅立叶变换,得到其时间轴上的有效不变域,将水印信号嵌入在这个不变域的中频部分以确保了水印的鲁棒性.在水印提取阶段,通过对水印的多重校验以确保水印信号的可靠性.实验结果表明,该算法对大部分的视频攻击都具有良好的鲁棒性.For the copyright protection of digital video, a novel video watermarking algorithm based on MPEG-7 contour description is proposed. The original video is spitted into a series of frame groups, for each frame in the group, the points of contour shape are described by MPEG-7, and taking advantage of hash function to calculate the keys between those points in the 1D-DFT transform domain and the watermark signals. Then, decomposes all those contour points of each frame in one group through 2D-DFT transform and obtains the invariant domain of the time axis and the contours, and then embeds the watermark signals into this domain to ensure the watermark reliability and robustness. And in the process of watermark extracting, it is performed multiple watermark verification to keep it accuracy. 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 傅立叶变换 MPEG-7 

分 类 号：TP309.1[自动化与计算机技术—计算机系统结构]