ZHANG Xu,LI Zhi-hui,WANG Xin,PENG Si-long,XU Xiao-jing,WANG Shi-jun
Abstract (
)
Download PDF
HTML
Knowledge map
Save
Digital videos and photographs can be altered rather easily since the broad availability of tools for the acquisition and processing in recent years, and using tampered images and videos to extort and threat becomes more and more serious. All these reasons raise the need to verify whether a multimedia content, which can be acquired by a video surveillance system, downloaded from the internet, or received by a digital TV broadcaster, is original or not. Previous work mainly focuses on image forensics, and it is only in recent years that forensic experts have begun specific research into video forensics. Compared with image forensics, video forensics have new challenges such as computing power because of the large amount and the complexity of data, and high compression factor; however, video signals have distinctive characteristics such as encoded mode, spatial and temporal characteristics, and ways of tampering giving video forensics a more abundant contents and broader application prospects. Signal processing experts have been investigating video forensic strategies and have made some progress, using the peculiarities of video signals and footprints left by alterations. Previous overview papers mainly address image forensic technology and only a few details are provided about video forensic analysis, so we present an overview of the video forensic techniques that have been proposed in the literature in this paper. Considering each frame as single image and many of image forensic tools can be applied to video signals as well, we first give a preliminary review of image forensics which provides the foundations for analogous techniques targeting video content in Section 1, such as camera artifacts, compression and geometric/physical inconsistencies. Next, we give a survey of video forensic technology focusing on various aspects related to video forensics such as acquisition, compression, and editing operations, and summarize the strengths and weaknesses of each solution. Then we address video acquisition in Section 2, presenting several strategies to identify the device that captured a given video content. Then, we consider the traces left by video coding, which are used to determine, e.g., coding parameters, coding standard, number of multiple compression steps or network footprints in Section 3. Video doctoring is addressed in Section 4, which presents forensic analysis methods based on detecting inconsistencies in acquisition and coding-based footprints, as well as methods that reveal traces left by the forgery itself such as inconsistencies in content and copy-move detection in video. As the design of novel forensic strategies aimed at image and video paralleled by the investigation of corresponding anti-forensic methods, we provide a brief introduction to anti-forensic techniques in Section 5. Video forensics is becoming research focus gradually, receiving more and more attention, and still presents many unexplored issues that wait for further study and deeper exploration.