| Summary: | 博士 === 國立臺灣科技大學 === 資訊工程系 === 102 === The tremendous advances in digital media content management and distribution tools and ever evolving social networks, blogs, and video sharing websites all contributed in making it rather easy to access, edit and redistribute digital multimedia contents such as images, video, text, and audio. Thus, a pressing need arises to protect multimedia data from being misused or its digital copyrights breached through various methods of illicit content manipulation and piracy.
Digital watermarking is providing a promising method of protecting digital data from illicit copying and manipulation by embedding a secret code directly into the data. Digital watermarking allows a user to add a layer of protection to the digital media content by identifying copyright ownership and delivering a tracking capability. That addition monitors and reports where the user’s digital media contents are being used enabling the enforcement of proper distribution channels maintaining the established rights of original authors and owners.
This dissertation aims at presenting advanced digital watermarking algorithms for two different digital media contents, namely relational databases and H.264/AVC compressed videos, for copyright protection and content-authentication purposes. We find these two domains are of utmost importance to focus on currently since these types of media lack copyright protection and content protection techniques.
Pertaining to the relational database watermarking, we present three algorithms. First one is a blind and robust distortion-free algorithm. The second one is a blind and robust reversible algorithm dedicated to applications which don’t permit permanent errors. And the third one is a new authentication protocol established to solve the problem of falsely claimed ownership by introducing a trusted time-stamping service (TSS) to the watermarking method which plays a significant role in authenticating the watermarked relation, the embedded watermark and the relation’s owner.
In the field of H.264/AVC video watermarking, two algorithms are introduced to tackle the content-authentication problem based on fragile watermarking. In the first algorithm, the H.264 variant features are securely hashed and embedded into IPCM-block low pixel samples. In the second algorithm, the features are securely hashed and embedded into the last nonzero ac residual of some carefully selected luma 4×4 sub-blocks to achieve the highest levels of sensitivity to intentional and unintentional modifications while maintaining the visual perception and video coding efficiency. Whereas the other algorithm is based on semi-fragile watermarking scheme in which tamper detection and localization of attacks is achieved.
As a result, various aspects of digital watermarking such as imperceptibility, robustness, fragility, semi-fragility and reversibility each best suited to the media content have been proposed and reported. Experimental results we obtained show the effectiveness of the proposed methods in thwarting and/or detecting several types of attacks while maintaining high perceptual quality, reasonable payload capacity and coding efficiency of digital media.
|
|---|
