Double H.264 Compression Detection Scheme Based on Prediction Residual of Background Regions

摘要: Detection of double video compression plays an important role in forensics. However, existing methods rarely focused on H.264 videos and are unreliable to provide detection results for static-background with fast moving foregrounds. In this paper, effective scheme based Prediction Residual Background Regions (PRBR) is proposed overcome these limitations. Firstly, the mask background regions each frame obtained by applying Visual Extractor (VIBE). VIBE efficient robust subtraction algorithm, which can distinguish foreground at pixel level. Then, PRBR feature designed characterize statistical distribution average prediction residual within mask. After that, Jesen-Shannon Divergence introduced measure difference between features adjacent two frames. Finally, a periodic analysis method applied final sequence estimation first Group Of Pictures (GOP). Eighteen standard testing sequences captured fixed cameras used establish dataset. Experiments demonstrate achieve better performance compared the-state-of-art methods.