采用光流场运动估计的双树复小波域视频水印算法

A Digital Watermarking Algorithm Using the Motion Estimation of Optical Flow Field in the DT-CWT

提出一种基于光流场的双树复小波变换(DT-CWT)的数字水印算法.首先利用光流信息检测出视频序列的关键帧,然后利用光流场技术对选取的关键帧进行局部区域的运动估计.最后,根据人眼对快速流动物体敏感性弱的特点,将水印信号分块嵌入关键帧内快速运动区域.测试结果表明:该算法具有很好的不可见特性,在面对噪声、缩放等恶意攻击时也具有很好的稳定性,同时也具有很好的实时性.

This article presents a digital watermark algorithm based on optical flow field in dual tree complex wavelet transform （DT-CWT）. In the beginning, detected the key frame of the video by optical flow information, then do the motion estimation of the key frame using optical flow field. According to the sensitivity of the human eye for the rapid flow things, embed the sub-block watermark in the area of rapid movement. The measuring result shows that this algorithm has great features of the sight. At the same time, itrs has great stability face to the noise, scaling and other malicious at- tacks, it also shows a good real-time.