Digital multimedia forensics techniques for digital video content protection

Abstract

With highly sophisticated digital imaging technology, the acquisition and utilization of high-quality digital image contents are common in our life. The benefits of digital image contents, such as the ease of editing, sharing, and storing, have led people to use digital image contents more common. Furthermore, analog image contents are replaced by digital image contents in many areas such as journalism, court, medical science, contest, and so on. Digital image contents come into wide use due to their great benefits, however, increase in digital image contents use brought many misuses. The most critical abuse is forgery on the fact captured in the digital image contents. Since digital image contents inherits the very nature of digital data, which is ease of manipulation, the authenticity of digital image contents is questioned. Furthermore, the appearance of the digital contents editing tools, such as Photoshop, Illustrator, Movie maker, and Premiere, makes ordinary people possible to forge the digital contents without any professional knowledge. It can help to create fake transcripts, counterfeits, fake certifications, and other forged digital images and videos. These days, various forged image contents spread out and used to get a job, earn money, and gain social promotion in society. To authenticate the integrity of digital image contents, various passive multimedia forensics techniques were investigated for last few decades. However, only little attention was focused on advancing the techniques to detect various video tampering attacks. Therefore, this dissertation proposes forensics techniques as countermeasures for the protection of digital video contents.

First, the dissertation presents a forensic technique that detects the frame-rate converted video. Video frame rate conversion is one of the common temporal operations in video tampering. When multiple original video clips are used to create a forged video, each original video clip is often acquired at different frame rate. Therefore, the frame rate of each original video clip needs to be united to create a combined video clip. Furthermore, the frame rate conversion technique can be used to convert the video playback speed while the frame rate is conserved. It desynchronizes the video watermark that utilizes the temporal information. Up to now, to the best our knowledge, only a few work for the detection of the frame rate conversion have been presented. The conventional temporal interpolation methods uses only pixel information between consecutive frames to create interpolated frames. Also, advanced temporal interpolation methods uses pixel information and motion information between consecutive frames to create interpolated frames. These methods leaves periodic artifact on motion vectors in the frame-rate converted video. Therefore, we exploited the periodic properties of temporal interpolation by the second-derivative of the position in the frame-rate converted video. The motion vectors between consecutive frames were estimated and validated to analyze the periodicity accurately. Furthermore, the original frame-rate was also estimated. Empirical evidence from a large database of frame-rate converted videos, which were converted using nearest neighbor method, bilinear method, and motion-based method, indicates the superior performance of the proposed method.

Second, the dissertation presents a forensic technique that detects the recaptured video based on sensor pattern noise (SPN). With the advances in digital camcorders, video recapturing is becoming a significant problem. The most significant problem of pirated movies recaptured by camcorders is that the illegal copies appear on the Internet just a few hours after a film's release and before the legal DVDs become available. Only one camcorder is needed to copy a movie and to trigger massive unauthorized reproductions and distribution of millions of illegal copies of the movie. The annual amount of revenue loss caused by these illegal copies has reached US$ 6.1 billion, and approximately 90% of these illegal copies are pirated via video recapturing. Nevertheless, little research on recaptured video detection has been undertaken. Therefore, we propose an automated method for the detection of recaptured videos. For this purpose, the recapturing process was modeled and the differences between the production processes of the original videos and recaptured videos are analyzed. From the analyzed differences, SPN and shot-based detection method was proposed. Moreover, the SPN merging and high frequency map are derived in order to overcome the low quality of the shot-based SPN. Empirical evidence from a large database of test video, including compressed (Q60~Q100), scaled (0.6~1.0), and cropped video (1.0~0.6) videos, indicated that superior performance of the proposed method.