摘要
提出一种在MPEG-2视频流中嵌入水印的方法。将水印嵌入到所有B帧的运动矢量中,嵌入时对每帧中的运动矢量进行分组,每组嵌入1位水印信息,提高水印的鲁棒性。采用一种自适应策略,通过修改运动矢量幅值的奇偶性来嵌入水印,减少对运动矢量个数及其幅值大小的改变,提高水印在压缩视频中的不可见性。实验结果表明,该算法具有较好的抗多次编码及码率改变攻击的能力,能够满足水印的实时嵌入和提取。
This paper proposes a blind watermark method embedded in MPEG-2 video bitstream. The watermark is embedded in motion vectors of all B frames which are grouped, every group embeds one bit watermark. By adopting an adaptive strategy and modifying the parity of the value of motion vector, the modification of the number and the value of motion vector are reduced, it improves the imperceptive effect in the compressed video. Experimental result shows that the algorithm is fairly resistant to multi-time coding and bitrate changed attack, at the same time, the proposed algorithm can satisfy real-time embedded and extracting watermark.
出处
《计算机工程》
CAS
CSCD
北大核心
2010年第6期142-145,共4页
Computer Engineering
基金
山东省自然科学基金资助项目(Y2008G11)
青岛经济技术开发区科技发展计划基金资助项目(2007-2-37)
山东科技大学科学发展基金资助项目(05g018)
关键词
视频水印
自适应算法
运动矢量
对称性
video watermark
adaptive algorithm
motion vector
symmetry