视频数据隐藏的差分能量调制新算法

Robust Video Concealment Based on Differential Energy Modulation

提出了基于差分能量调制(DEM)的盲视频水印算法。针对差分能量水印(DEW)算法可能造成视觉失真严重、水印容量小和鲁棒性较差的不足,提出用差分能量信号的调制值来承载隐藏信息位的新方法。通过对量化块中水平子区域与竖直子区域之间的能量差值实施量化索引调制(QIM),差分能量值能被微调至最近邻的对应水印位量化区域;同时由于调制前后的能量误差被分配到子区域中幅值较大的10个系数上,并对调制前后的码字替换进行了相似度选择,嵌入信号因而能很好地保持原信号的视觉质量和码序列长度。实验结果表明,DEM算法在水印的视觉失真与鲁棒性能、水印容量以及码率保持性能方面都好于DEW算法;另一方面,DEM算法是在一个区域中嵌入水印,所以它的鲁棒性能又比只在某个系数上嵌入水印的QIM算法好。

To deal with the drawbacks of differential energy watermarking（DEW） algorithm such as its poor transparence, low capacity and weak robustness, we propose a blind video watermarking scheme related to differential energy modulation （DEM）. It chooses the embedded regions of a quantized block, and labels a hiding bit by dithering differential energy between the horizontal and the vertical sub-regions of that block, with an adaptive quantized index modulation （QIM） method. To assure transparence and robustness, the embedded error is spread over 10 biggest coefficients of each host block. The perceptual quality and the length of an original sequence could also be kept unchanged owing to its similitude codeword replacement. Experiments demonstrate that DEM is superior to DEW in the performance of transparence, capacity, robustness and rate maintenance. At the same time it achieves a better robustness than QIM, for QIM carries one bit in one coefficient, whereas DEM embeds one bit in one region.