【作者】 庞磊；

【导师】 吴一全；

【作者基本信息】 南京航空航天大学 ， 信号与信息处理， 2009， 硕士

【摘要】 近年来数字视频产品的大量出现,使市场对其版权保护技术的需求更加迫切。但是,现存的数字水印研究文献中,图像水印远多于视频水印,因而视频水印技术的研究应获得更多的关注。本文以原始视频作为研究对象,研究并实现了基于Contourlet和独立分量分析的视频水印算法。首先,实现了一种基于Contourlet变换的能够有效抵抗几何形变攻击的视频水印算法。运用平均交流能量(AAE),并根据Contourlet变换的空—频特性嵌入水印。结果表明,该算法在保证视频视觉质量的同时具有很强的抵抗几何形变攻击的能力,对于低通滤波等其他常规攻击也具有鲁棒性。然后,通过采用空域的Contourlet变换和时域的一维小波变换,给出了一种新型三维多分辨率分解方法,并将水印信息嵌入到视频三维多分辨率分解后空域子带系数的静止成分和动态成分中。结果表明,该算法具有很强的鲁棒性,并且能取得良好的视觉效果。接着,提出了一种基于背景的能够抵抗几何攻击的盲视频水印算法。利用独立分量分析(ICA)提取镜头中一个视频段的背景图像,并将水印嵌入到背景图像的低频子带中。通过检测含有水印背景和不含水印背景的低频子带之间的差异来提取水印。结果表明,该水印算法具有很强的抵抗几何形变攻击的能力,对于其他常见攻击也具有一定的鲁棒性。最后,研究了一种基于Contourlet域独立分量分析的视频水印算法。对同一镜头中的视频帧分别进行Contourlet域的独立分量分析,并将水印嵌入到独立分量分析所得的运动分量的带通子带上。结果表明,该算法能够有效地抵抗MPEG压缩及其他常见攻击,而且还大大减少了计算量。更多还原

【Abstract】 In recent years, the increasing of digital video product makes a great demand of copyright protection technology. However, among the existing digital watermarking literatures, image watermarking algorithms are far more than video’s. So more attentions should be paid to video watermarking algorithms. This paper takes original videoes as research object and video watermarking methods based on contourlet transform and independent component analysis(ICA) are researched and realized.Firstly, a video watermarking scheme resistant to geometry transformation attacks is realized. In this scheme, average AC energy(AAE) is used for the embedding of watermark according to the characteristic of spatial-frequency of contourlet transform. Experimental results show that the proposed scheme can make the watermark resist geometry transformation attacks effectively and keeps the visual quality of the video. In addition, it also has enough robustness to resist other attacks, such as lowpass filtering attacks.Then, a new 3-D multi-resolution transform based on contourlet transform in the dimensional domain and 1-D DWT in the temporal domain is given. With this kind of 3-D multi-resolution transform, watermark can be embedded into the quiescent component and the dynamic component of the sub-band in contourlet domain. The experimental results show that, this algorithm has enough robustness and the watermarking is invisible.Next, a video blind watermarking algorithm resistant to geometry transformation attacks is proposed. In this scheme, the background of a video segment in the video shot is extracted by ICA, and meaningful watermark is embedded into the lowpass subband of the background. The watermark is extracted by detecting the distinction of the lowpass subbands of the background image with watermark information and the background image without watermark information. Experimental results show that, this algorithm can make the watermark resist geometry transformation attacks effectively and keeps the visual quality of the video. In addition, it also has enough robustness to resist other attacks.Finally, a video watermark scheme based on ICA in contourlet domain is researched. Frames in the same video shot are analyzed by ICA in the contourlet domain. Then, the watermark is embedded into the contourlet coefficients of the motion component. Simulation results show that the proposed scheme can preserve high video quality. The watermark can be extracted blindly and is robust to MPEG compression and other common video processing. Further more, it is simple with much less computation amount.更多还原