| With the rapid development and wide applications of multimedia and networktechnology, the protection for multimedia contents had become the most urgentproblem to be solved. Digital image is a very important kind of multimediainformation, the protection for digital image content can be divided into two aspects:copyright protection and content authentication. The traditional encryptiontechnology protected the digital image content with large limitation. However, thenew technology——digital watermarking technology can remedy these defects,which is a strong tool for multimedia content protection following the encryptiontechnology. Therefore, in this thesis, the digital image watermarking technologies forcopyright protection and content authentication of multimedia data in network aremade a deep study, and the maximum watermark capacity estimation methods forJPEG and JPEG 2000 watermarking models are proposed theoretically, and themodel of the illegal copy protection and real-time copyright tracking and diction fornetwork multimedia dada is proposed. The main researches in this thesis are asfollows:1. For robust digital watermarking technolgoy, based on the researches of theexisting algorithm, a new robust digital watermarking algorithm based DWT andSVD is proposed. At first, the host image is divided into two parts, and the two partsof the host image are analogous to two communication channels for transmission thewatermark signal. Then, using redundancy information between adjoining pixels ofthe watermark signal, the watermark signal is coded by multiple description codertechnology to obtain two descriptions of low frequency subband of the watermarksignal. At last, the two descriptions are embedded into the DWT domain and SVDdomain of the two parts of the host image respectively. The proposed algorithm codesthe watermark signal by multiple coder technology in order to achieve the bettercapability of correcting error at the terminal decoder. In this thesis, the robustnessagainst several kinds of image attacks are tested, including median filter, Gaussianfilter, image cut, image shift and JPEG and JPEG 2000 image compression. Theexperiment results show that the proposed algorithm has good robustness against thecommon image processing.2. For semi-fragile digital watermarking technology, a new semi-fragile digitalwatermarking algorithm based on image content is proposed for tamper localization and recovery. In this thesis, the two invariant properties of DCT coefficients in theprocess of JPEG quantization are used to generate and embed the watermark signal.There are two watermark signals: one is authentication watermark for image tamperlocalization, the other is recovery watermark for effectively recovering the tamperedimage, thereinto, the authentication watermark is generated from DCT low frequencycoefficients in the pairs of image blocks in order to effectively resist vectorquantization attack. In the process of authentication, the image block isself-embedded, that is to say, the authentication watermark can be extracted from thesame image block so as to improve the precision of tamper localization. The recoverywatermark is embedded into the other image block with Torus automorphismmapping technology and the secret key so as to guarantee the security and goodrecovery effect of the proposed algorithm. The experiment results show that theproposed algorithm has higher precision of localizing tampered image and bettereffect of recovering tampered image, and effectively resist JPEG loss compressionwith certain quality factor, and reject malice tamper and beat collusion attackeffectively.3. For estimation on the capacity of digital watermark, the existing methods forwatermark capacity estimation are analyzed. Based on the drawbacks in the existingmethods, the watermark capacity estimation methods for JPEG and JPEG 2000images are proposed with Watson human visual model. For the JPEG watermarkingsystem, the maximum watermark capacity formula for computing JPEG image isobtained according to modified Just Notice Difference (JND) value. Because theJPEG2000 is a new image coder international standard, and the quality of JPEG2000compression is better than JPEG compression image obviously, so the estimationmethod on the watermark capacity for JPEG2000 watermarking system is furtherstudied. The maximum watermark capacity formula for computing JPEG2000 isobtained. The experiment results show that the total embedded watermarking bitscomputed by the proposed methods can satisfy the requirement of invisibility.4. For real-time copyright tracking and detection, the illegal copy protection andcopyright tracking and detection for multimedia data based on mobile agenttechnology is proposed, which is fit for copyright protection and tracking anddetection of multimedia data in the large distributed network.. Contents offering stage,verification stage of contents illegally copied, the phase of creating and dispatching amobile agent and the phase of detecting the watermark based on mobile watermarkare analyzed in detail. Comparison with the existing systems, the proposed systemdoesn't need watermark detector or smartcard but embeds an agent code into the host digital content. When the illegal copy behaviors happened in the user's host, thehidden agent will automatically execute its code to terminate the malice behaviorsarbitrarily so as to the purpose of the illegal copy protection. For the real-timewatermark detection, a watermark agent centre is designed to create and dispatch amobile watermark agent, and made the mobile agent execute its tasks, self-coping andself-ending. Every sub-model in the proposed system is independent, which is easy tobe expanded.With the development of computer, network and information science andencryption, secure and reliable watermarking technology will be more widely anddeeply used in all kinds of areas such as electronic commerce, network multimediaservice, remote diagnosis and so on.There are a great deal of hard work to becontinued on above theories and the application and development of the keytechnologies.
|
PDF Full Text Request