Abstract: This paper proposes a video heterologous tampering detection algorithm which is based on compressive sensing，considering the characters of the intrinsic noise in the video sequence.Pick up the noise information of each frame and establish the noise matrix in the video，compress them through the compressive sensing theory，diminish the redundancy rate for the noise information of each frame image tremendously，employ cosine similarity to the compressive noise matrix，get the similar matrix between the frame images and construct the tamper measure，apply the parameter model to detect the images which are tampered heterologously in the video.The experiment demonstrates that the proposed algorithm can detect the position of heterologous tampering effectively at a low compression rate in the video sequence，and obtain the higher accuracy of tampering detection than two existing algorithms.

Key words: noise information, compressive sensing, cosine similarity, video tampering detection

摘要： 考虑到视频序列固有噪声特征的特点，提出一种基于压缩感知的视频异源篡改检测算法。提取视频中每帧图像的噪声信息并建立噪声矩阵，通过引入压缩感知理论对噪声矩阵进行压缩，极大地降低每帧图像噪声信息的冗余度，对压缩噪声矩阵使用[cos]相似性衡量，得到帧图像间的相似度矩阵，并构造篡改度量，利用参数模型对视频的异源篡改图像进行检测。实验表明提出算法能以较小的压缩比对视频序列中的异源篡改位置进行有效检测，并得到比现有两种算法更高的篡改检测准确率。

关键词: 噪声信息, 压缩感知, cos相似度, 视频篡改检测