基于P、B帧MBNV特征的视频篡改检测方法

Video forgery detection based on MBNV feature in P & B frames

随着视频编辑软件的普及使用,视频篡改变得越来越容易。文中针对视频帧间篡改行为,分析讨论了视频帧中宏块数量在视频篡改情况下的变化规律,提出了基于视频P、B帧宏块数量变化(MBNV)特征的视频篡改检测算法,与现有算法不同的是:文中算法考虑了编码中常见的B帧在经历帧间篡改操作后的异常特性。算法通过对异常帧的判断提取,利用I帧的周期性特性,判断视频是否经过篡改,并对视频的篡改手段进行鉴别。实验证明,文中提出的方法对不同的视频编码方式和不同的比特率都有不错的检测效果。

With the popularity of the video editing software, tampering on video becomes easier andeasier. In order to detect inter-frame forgery, this paper discusses and analyses how the numbers ofdifferent kind of macroblocks change after the forgery and proposes an algorithm based on MacroBlocksNumber Variation （MBNV） of P , B frames, the advantage of the proposed method is considering theinfluence of B frames which are very common in videos. The method extracts the abnormal frames fromthe video, and uses the periodic property of I frames to detect whether a video is tampered or not. To thetampered videos, it also identifies the forgery types and locates the manipulated positions in forgedvideos. The experiments show the effectiveness of the algorithm.