计算机应用 ›› 2011, Vol. 31 ›› Issue (01): 258-259.

• 模式识别 • 上一篇    下一篇

基于运动目标检测的视频水印算法研究

陈希,周萍   

  1. 桂林电子科技大学
  • 收稿日期:2010-06-09 修回日期:2010-07-20 发布日期:2011-01-12 出版日期:2011-01-01
  • 通讯作者: 陈希
  • 基金资助:
    国家自然科学基金资助项目

Video watermarking based on moving object detection

  • Received:2010-06-09 Revised:2010-07-20 Online:2011-01-12 Published:2011-01-01
  • Contact: Chen Xi

摘要: 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。

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

Abstract: Abstract:To improve the robustness of video watermarking, a video watermarking algrithom based on moving object detection was proposed. The paper maked use of the temporal differecing algrithom 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, calculating the peak signal-to-noise ratio to show that this scheme has a great invisibility and concealment, and using the strimark software to geometric attack on watermarked image and analysising the correlation coefficient to prove that this algorithm has a great robustness.

Key words: Moving object detection, Singular value decomposition, Temporal differecing, Peak Signal-to-Noise Ratio, Video watermarking