一种用于帧内篡改检测的压缩感知视频水印实现算法

A Video Watermarking Algorithm for Intraframe Tampering Detection Based Compressed Sensing

针对视频水印在帧内篡改检测方面定位精度的不足,通过压缩感知对MPEG-4(Moving Picture ExpertsGroup-4)视频内容的特征表示,提出一种新的视频水印生成方法及其帧内篡改检测算法.该算法由压缩感知DCT(DiscreteCosine Transform)测量矩阵对I-VOP(Intra-Video Object Plane)图像提取U、V特征参数,生成基于内容的压缩感知视频水印数据并嵌入到图像Y分量的DCT中高频系数中实现帧内篡改检测.实验结果表明,与Hash视频水印算法比较,压缩感知视频水印数据具有更好的恢复能力,且水印算法对视频帧内篡改定位精度更高.

In order to overcome the drawback of localization accuracy for intraframe tampering detection based on video watermarking, this paper proposes a novel video watermarking generation and intraframe tampering detection algorithm based on feature representation of MPEG-4（Moving Picture Experts Group-4） video content by compressed sensing（CS）.First,U and V,feature parameters of I-VOP（Intra-Video Object Plane） frame,are extracted utilizing projection of DCT（Discrete Cosine Transform） measurement matrix of the CS,generate the CS-watermarking of the video content,and embed the CS-watermarking data into the coefficients of low frequency and high frequency of DCT domain in the Y component of image.Compared to Hash-watermarking generating algorithm,experiment results show that the CS-watermarking algorithm has better data recovery capability and localization accuracy for video intraframe tampering detection.