Video Forgery Detection Based On Macroblock Feature

With the development of video encoding and decoding technology and the popularity of video editing software,videos have enriched people's daily life.The tampered videos easily bring a lot of problems related to personal privacy,judic ial forens ics,social stability and other issues,at the same time as it becomes more and more easy to tamper a video,the research about video forgery detection has important practical significance.From the perspective of the inter-frame forgery in video,This paper analyses and discusses the feature about the variation of macroblock types in tampered videos,and puts forward a forgery detection method based on the feature of macroblock.The work and innovation of this paper have the following points:1.This paper proposes a feature based on macroblock types to describe forgery videos.The similarity of the video frames is related to the number of macroblock type.In general,forgery will weaken the frame correlation,which can be reflected in the variation of the corresponding macroblock.Therefore,we can use the number of macroblock types to represent the characteristics of forgery videos.2.This paper proposes a forgery detection algorithm based on the variation feature of macroblock numbers(MBNV).The specific princ iple is that it has macroblock variation anomalies when the tampered frame or the frame which is coded as I frame in the first compression are re-encoded as P or B frames.More Specifically,the number of I-MB increases and the number of S-MB decreases in P frames,the percent of forward prediction macroblock changes rapidly in B frames.We can detect forgery using the above characteristics.For the forgery video,this paper checks the autocorrelation of the periodic analys is to determine the type of video forgery,and locate the video forgery3.This paper designs and implements a inter-frame forgery detection system based on macroblock feature.The system includes preprocessing,video forgery detection,detection results output.It provides the users single video detection,batch detection,deriving detection results functions.