基于压缩感知的视频双水印算法研究

针对数字视频的内容保护与帧内、帧间篡改检测的难题,采用压缩感知理论提取视频的内容特征作为水印,提出一种双水印的视频保护和篡改检测算法。首先,利用压缩感知过程提取I帧宏块的内容特征,生成半脆弱的内容认证水印；然后,对帧序号进行二值运算,生成完整性水印；最后,利用压缩感知信号重构OMP(Orthogonal Matching Pursuit)算法把生成的双水印嵌入到I帧和P帧相应宏块的DCT高频系数的压缩测量值中,以此提高视频水印的抗攻击能力,并实现对视频篡改的检测。仿真实验表明,所提算法对视频帧内篡改具有精确定位到子块的检测能力；同时对帧插入、帧删除、帧交换等类型的视频帧间篡改具有很强的检测能力。

Double Video Watermarking Algorithm Based on Compressive Sensing

For the problem of video content protection and tamper detection in frame or between frames,by extracting the video content features as watermarking based on compressive sensing(CS) theory,we proposed a video protection and tamper detection algorithm based on double watermarking.Firstly,the content features of the macro blocks of I frame are extracted by CS procedure,and the semi-fragile authentic watermark is generated.Then,by performing binary arithmetic to the frame sequences,the integrity watermark is generated.Finally,the double watermarking is embedded into the CS measurements of high-frequency DCT coefficients in macro blocks of I frame and P frame by the OMP procedure of CS reconstruction.This algorithm can improve the anti-attack capability of video watermarking,and detect the tamper to video.Simulation results show that the proposed algorithm has accurate detecting capability for tamper in frame,which can detect the sub-blocks of video frame.Also,it has precise detecting capability for tamper between frames such as frame insertion,frame deletion and frame switch,which can detect any tampered frame.