一种基于时空冗余统计不可见性的视频隐写分析方法

视频中的扩频嵌入和MSU隐写是两种典型的隐写方法,可抵抗压缩等多种攻击。为有效检测利用此方法嵌入的秘密信息,根据视频的时空冗余特性,提出了一种实时的视频隐写分析方法。用大小为L+1的滑行窗口获取视频帧估计值,并提取相应的DCT和马尔可夫特征,使用神经网络、支持向量机等多种分类方法对隐写视频进行检测。结果表明,根据DCT和马尔可夫特征来分析,检测率较高。将支持向量机和时、空冗余特性等应用于视频隐写分析,有很大前景。

Video steganalysis method based on spatial-timeredundancy of statistics invisibility

The video steganography of the spread spectrum embedding and MSU are the two typical steganographic method , which can be resistant against compression and other attacks, in order to obtain the effective detection of the embedded secret information. According to the video spatial-time redundancy, this paper proposed a real-time video steganalysis method. Used a size of L+1 sliding window to obtain an estimate of the video frames, and extracted the corresponding DCT and Markov features, and used neural networks, support vector machines and other classification methods for video steganalysis. The results show that, according to the DCT and Markov features, the correct detection rate is higher. The support vector machines and time and spatial redundancy, etc can be used in the video steganalysis, which have great prospects.