Digital Video Passive Authentication For Temporal And Spatial Tampering

With the rapid development of the network and digital multimedia technology, the transmission of multimedia files, such as images, audios, videos, have become more efficient. The appearance of high-quality imaging equipment and the progress of multimedia editing software also made editing forgery much easier. While bringing lots of convenience to us, it also causes lots of problems. The malicious tampered videos may carry false information, spread fast through the network and confuse the public. Some criminals create fake evidence by tampering with videos, which undoubtedly have a certain impact on social stability. Therefore, the video forgery detection technology has become a hot spot research in the field of information security.At present, most digital multimedia forensics research focus on digital images. Compared with the digital image, digital video is more complex, because a video sequence has larger data volumes and higher dimensional space. The technology for video forgery detection is not universal at home and abroad. The methods of video forgery can be divided into two groups: inter-frame forgery and intra-frame forgery. Inter-frame forgery includes frame insertion, frame deletion, frame replacement and frame copy-move. Intra-frame forgery focuses on tampering with the video frame content, object deletion and matting synthesis are the two common types. For detecting the video inter-frame forgery and video intra-frame forgery, the main work of this paper has following aspects:(1) A method for the video inter-frame forgery, includes frame insertion, frame deletion, frame replacement and frame copy-move. This method is a double checking mechanism based on Zernike opponent chromaticity moments and coarseness feature analysis by matching from coarse to fine, it can not only detect the tampering location accurately, but also improve the time efficiency greatly.(2) A method for the video moving target deletion forgery, which is based on energy suspicious degree calculation. This method can effectively detect whether the moving target was removed and orient the temporal and spatial tampering location accurately.(3) A method for Blue Screen Matting and Compositing, which is based on 3-Stage Foreground Analysis and Tracking. This method uses Gaussian Mixed Model, similarity analysis and Compressive Tracking algorithm, has higher accuracy and performs well in terms of speed.