Complete Video Quality Preserving Data Hiding for Multimedia Indexing


Digital video has become a popular multimedia content in both online and offline environments thanks to the advancement of computer and portable devices, as well as broadband internet technologies. While digital video is still gaining popularity, it is important to consider ways to protect the contents from malicious use, efficient ways to search the desired contents in the database, secure ways to provide extra features for upgraded viewers, reliable ways to recover from transmission error for uninterrupted viewing, etc., for the future of digital video. To accomplish such tasks, data hiding is one of the fields that provides the solutions (Johnson et al., 2003; Katzenbeisser & Petitcolas, 2000). In general, there are two types of data hiding for video: one that hides the video content itself (video encryption or scrambling) so that nobody understands what is being transmitted (Takayama et al., 2006; Wong et al., 2003; Zeng & Lei, 1999); the other that embeds external information into the video, hence utilizing video as the data host. We consider the latter in this chapter. Under this category, one of the basic requirements for a data hiding method is the ability to produce video of high image quality. On top of that, additional properties are desired, depending on the application in question. In case of watermarking, the information embedded into a video should be able to withstand some common image processing attacks such as re-compression at different bitrate, random video frame dropping, resizing, etc. (Cox et al., 2002). In case of steganography, the embedded information should stay undetectable with respect to steganalysis (Budhia et al., 2006), which is a process for revealing the existence of hidden information in a suspicious video. In applications such as annotation or indexing, even though it is not a compulsory property, it is usually preferable to achieve reversibility so that the embedded information could be removed to restore the original video. Other applications of data hiding could be found at (Katzenbeisser & Petitcolas, 2000; Kurosaki & Kiya, 2002; Yanagihara et al., 2005). Some representative data hiding methods in video domain could be found at (Bodo et al., 2004; Kiya et al., 1999; Liu et al., 2004; Nakajima et al., 2005; Ni et al., 2006; Qiu et al., 2004; Sarkar et al., 2007; Xu et al., 2006; Zhang et al., 2001). For example, Kiya et al. embed information into an MPEG compressed video by modifying coefficients at selected location(s) in each 8× 8 qDCTCs (quantized DCT coefficients) block (Kiya et al., 1999). The quantization table is further modified to suppress distortion. Nakajima et al. proposed a high carrier capacity data hiding method utilizing the idea of zerorun length coding in MPEG domain (Nakajima et al., 2005). After zigzag scanning, a dummy nonzero value is inserted at a location that is

18 Figures and Tables

Cite this paper

@inproceedings{Wong2012CompleteVQ, title={Complete Video Quality Preserving Data Hiding for Multimedia Indexing}, author={KokSheik Wong and Kiyoshi Tanaka}, year={2012} }