【摘要】 为了增强抗攻击的能力和保证算法的不可见性,基于DC直流系数的稳定性,把选择的视频帧分成8×8块,每块经过2次DCT变换,提取与每块像素RGB平均值成正比的DC直流系数,将这些DC系数组织成一维序列,并划分为更小的子序列,然后对每个子序列进行一维DCT变换,通过比较相邻2个子序列中最后一个高频系数值的大小来嵌入水印.实验结果表明:该算法有较高的视觉质量,同时对于MPEG压缩,帧置换,加噪等处理有良好的鲁棒性,在水印的提取过程中不需要任何附加信息.更多还原

【Abstract】 Using the DC coefficient’s stability,a blind video watermarking method is proposed.In order to enhance the strength against the attacks and guarantee the scheme’s invisibility,some frames are selected to embed watermark.The frame is divided into many 8-by-8 blocks.With the block being transformed twice by DCT transformation,the DC coefficient can be extracted.All the blocks’ DC coefficients are organized into one dimension sequence and divided into smaller sequences,then one dimension DCT transformation is assigned to every sequence.The ultimate watermark is embedded by comparing the two adjacent sequences’ last high frequency coefficients.And the watermark is extracted directly from the bit stream independent of the original video frames.Experimental results show that this algorithm achieves high visual quality,and robust to MPEG compression,frame permutation and noise disturbance.更多还原