基于运动目标检测的视频水印算法  

Video watermarking algorithm based on moving object detection

作　　者：陈希[1] 周萍[2] 

机构地区：[1]桂林电子科技大学计算机科学与工程学院,广西桂林541004 [2]桂林电子科技大学电子工程与自动化学院,广西桂林541004

出　　处：《计算机应用》2011年第1期258-259,262,共3页journal of Computer Applications

基　　金：国家自然科学基金资助项目(60961002)

摘　　要：为了提高视频水印的鲁棒性,提出一种基于运动目标检测技术的算法。通过相邻帧差法提取并标记视频图像序列中的运动目标,并采用图像局部奇异值分解(SVD)算法,实现水印的嵌入和盲提取过程。在仿真实验中,通过计算水印嵌入后图像的峰值信噪比,证明该水印算法具有很好的不可见性和隐蔽性;并使用strimark软件对嵌入水印后图像进行几何攻击,分析水印图像的相关系数,验证本算法具有很好的鲁棒性。To improve the robustness of video watermarking, a video watermarking algorithm based on moving object detection was proposed. The paper made use of the temporal differencing algorithm to extract and sign the moving targets in the video image sequences, and then, achieved watermark embedding and extraction processes by Singular Value Decomposition （SVD） method. In the simulation, the Peak Signal-to-Noise Ratio （PSNR） was calculated to show that this scheme has a great invisibility and concealment, the strimark software was used to geometricly attack on watermarked image, the correlation coefficient was analyzed to prove that this algorithm has a great robustness.

关 键 词：运动目标检测 奇异值分解 相邻帧差法 峰值信噪比 视频水印 

分 类 号：TP309[自动化与计算机技术—计算机系统结构] TP391.41[自动化与计算机技术—计算机科学与技术]