基于视觉显著模型的音视频交叉水印算法

提出一种基于Itti视觉显著模型的音视频交叉水印算法.首先利用音频码流的低通幅值特性生成水印,并嵌入到视频码流中;其次解码视频码流的关键帧Ⅰ帧,获取YUV序列上的亮度(Y)分量,并对其进行显著性分析;最后分块对非显著区域进行DCT变换,并通过量化索引调制嵌入水印.实验结果表明,该算法将嵌入水印后的关键帧平均PSRN值提高了1.678 5 dB,缓解了水印不可见性和鲁棒性的矛盾,实现了音视频码流之间的交叉保护.

Audio and video cross watermarking algorithm based on visual saliency model

This paper proposed a simple audio and video cross watermarking algorithmbased on ltti's visual saliency model.Frist of all,it utilized the low-pass amplitude characteristic of audio streamingto generatewatermarking,embedded into the video streaming;Secondly,decoding the keyframe,Ⅰ frame,of video streaming to obtain the Y component of YUV sequence,remarkably analyzing it;Finally,non remarkable area is blocked into DCT,embedded intowatermarking by quantization index modulation.The experimental results show that the average PSNR of key frame,in this algorithm,is lifted by 1.678 5 dB,which relieved the contradiction of watermark invisibility and robustnessand achieved the cross-protection between the audio and video streams.