基于运动矢量多直方图修正的视频可逆隐写算法

针对基于直方图平移的视频隐写算法与视频内容相关性不强进行了研究，提出了一种基于运动矢量多直方图修正的H.264视频可逆隐写算法。该算法通过统计特定帧中宏块移动的速度特性和散乱度对人眼视觉的敏感程度设计修正参数，再根据修正参数设计多个二维直方图，最后在非特定帧中通过二维直方图修正算法嵌入信息。实验结果表明：与同类型视频隐写算法相比，嵌入容量平均提高8.2%，在满嵌时该算法的PSNR和SSIM的变化值分别平均降低19.9%和17.5%，该算法在保持较低的比特率变化下，具有更好的嵌入容量和不可感知性。

Video steganography algorithm based on motion vector and multi histogram modification

Aiming at the weak correlation between video steganography algorithm based on histogram translation and video content, this paper proposed a multiple histogram modification H. 264 video steganography algorithm based on motion vector. The algorithm designed correction parameters by counting the speed characteristics of macroblock movement and the sensitivity of dispersion to human vision in a specific frame, then designed multiple two-dimensional histograms according to the correction parameters, and finally embedded information in a non-specific frame through the two-dimensional histogram modification algorithm. The experimental results show that compared with the same type of video steganography algorithm, the embedding capacity is increased by 8.2% on average. When fully embedded, the change values of PSNR and SSIM of the algorithm are reduced by 19.9% and 17.5% respectively. The algorithm has better embedding capacity and imperceptibility while maintaining a low bit rate change.