摘要
关于版权保护优化问题,视频水印普遍存在不能抗共谋攻击的问题。针对嵌入强度和抵抗攻击能力的不足,在分析共谋攻击的数学模型的基础上,提出了一种利用镜头分割的视频水印方案。首先利用累加直方图对视频进行镜头分割,再以镜头为单位嵌入水印。嵌入水印时,依据视频的特征及其DCT系数的关系确定嵌入域,提取嵌入域的局部运动特征,并对特征进行全局统计收敛,得到特征系数作为水印嵌入强度,采用扩频方法嵌入水印。实验结果表明,水印方案能抵抗各类共谋攻击,可为版权保护提供有效的方法。
ABSTRACT：A mathematical model was established to solve the common collusion attacks in video watermark. From this model a watermark scheme based on video segmentation was proposed： cumulative histogram was used for video segmentation, furthermore, different watermarks were embedded into different segmentations. The image＇s visual fea- tures and DCT coefficients of sub - region were utilized to obtain the watermark embedding locations. Local motion features were derived from every location and adjusted by a statistics model which was used as embedding strength. The watermark was embedded by spread spectrum method. Experimental results show that the watermarked video can resist to collusion, frame insertion, frame removal, non -synchronous, and MPEG- 2compression attacks.
出处
《计算机仿真》
CSCD
北大核心
2013年第4期421-425,共5页
Computer Simulation
基金
华北电力大学青年教师科研基金(200811030)
关键词
视频水印
共谋攻击
镜头分割
时间轴攻击
离散余弦变换
Video watermark
Collusion attack
Video segmentation
Temporal attack
Discrete cosine transform
