Research And Implementationof The Compressed Video Forensic Platform Based On Stream Parameter Characteristics

As progress and development of video codec and editing,videos have been widely used.While the video tampering has become increasingly frequent,the tampering detection of video has become a new hot spot in the field of multimedia forensics.This paper analyzes the H.264 codec process,the characteristics of stream parameters and the impact of tampering,proposes tamper detection methods based on the characteristic of stream parameters,achieve the video tamper detection platform.The main innovative achievements of the thesis include the following.First,extraction of stream parameters in the decoding process was realized,the tampering and detection for videos were analyzed.Based the current tampering,noted the impact of tampering in the video stream parameters in the process of the video encoding,extraction of stream parameters was completed by the development of codec platform.Second,this paper proposed the tampering detection method based the video frame rate.The algorithm based on the principle of video codec,detect the tampered by the duplication of one frame.Experimental results show that this method in the detection accuracy than previous algorithms improved,and the efficiency is also better than existing algorithms.Third,this paper proposed a method to detection the frames insertion based on the encoded residual.By comparison test,the method in the detection accuracy rate and recall rate than the previous algorithm has improved greatly,and the efficiency increased.Forth,With the development of the codec platform,achieves the tamper detection platform,based on the method based on the characteristics from stream,the method based on the residual and stream information and the methods proposed formerly.Now a day's the digital video is usually stored in compressed format.The differences which could hardly be found by naked eye,could be recorded in the coding process,and reflect in the compressed digital streams.As the tampering attacks happened with the compression of digital video,the differences between the original videos and the tampered videos would been recorded in the compressed stream.This paper proposed two methods in order to identify the tampering in the surveillance digital videos,one is based on the characteristics from stream to identify the duplication of frame;the other is based on the residual and stream information to identify the insertion of frames.The paper also describes the building process of tamper detection platform,based on the methods proposed.