The research for tamper forensics on MPEG-2 video based on compressed sensing

In order to overcome the shortcoming of the current video tamper forensics, we propose a video tamper forensics algorithm based on compressive sensing in this paper. It regards MPEG-2 based compressed video as the research target and decompresses it partly. By conducting random projection on some DCT coefficients of I frame with compressed sensing, the video features of I-frame are extracted and embedded into the DCT coefficients of I frame as watermark. We extract features and watermarks from video for the tamper forensics, and then locate the tamper area on video by matching them. Simulation results show that the algorithm has better effect on video tamper forensics.