Robust Adaptive Digital Multimedia Watermarking Techniques For Copyright Protection

With the fast development of the Internet and multimedia technology since 1990s, the storage, ateration, copy and transmission of the digital works via computers and Internet is becoming pretty easier and used more frequently. This causes the secure communication and authentication on digital works to face severe challenges. Traditional cryptography cannot solve this problem very well. In recent years, digital watermark has been proposed as a solution to the problem of copyright protection of multimedia documents in networked environments. There are two important issues that watermarking algorithms need to address. First, watermarking schemes are required to provide trustworthy evidence for protecting rightful ownership. Second, good watermarking schemes should satisfy the requirement of robustness and resist distortions due to common image manipulations (such as filtering, compression, cropping, rotation, etc.). The work presented in this thesis is concerned with the design of robust digital image and video watermarking for copyright protection.Firstly, the dissertation gives an overview of watermark history, the concept of digital watermark technology including watermark requirements and applications, and classification of watermarking, typical watermark system. Finally, an overview of image and video watermarking techniques are given.The watermark robustness against common signal processing operation and geometric distortion is an important attribute of digital watermarking schemes. In order to achieve a good balance between transparency and robustness, one effective way is to design adaptive watermarking techniques by exploiting human visual characteristics of original covers. The main contributions are followings.In Chapter 3, we propose two robust adaptive image watermarking algorithms using visual models in discrete cosine transform (DCT) and discrete wavelet transform (DWT) domain. The first algorithm utilizes the Watson's visual model (JND) to determine the watermarking strength necessary to invisibly embed the watermark in the DCT block coefficients of the cover image by controlling the power of strength of JND. The second blind DWT based technique embeds the watermark by modifying the coefficients of mid–frequency detail sub-bands of wavelet sub-block, chosen with secret key. The visual model is designed to generate a Just Noticeable Difference mask (JND) by analyzing image characteristics such as textures and luminance of the cover image in the DWT domain. Since the secret is needed for both embedding and extraction of watermark, it is impossible for an unauthorized user to extract the embedded watermark. The experimental results show that both techniques give a good result under various image signal processing operations including slight geometric distortions.A blind and efficient digital image watermarking algorithm is presented in the Chapter 4, which is operated in discrete cosine transform (DCT) and singular value decomposition (SVD). The proposed algorithm combines both merits of discrete cosine transform (DCT) and singular value decomposition (SVD). The watermark is embedded into the host image by applying a quantization index modulation process on the biggest singular values of image blocks in DCT domain. We developed a visual model by taking into consideration of blocks statistics of the host image to avoid serious visual degradation of the watermarked image. Extensive experiments show that the proposed algorithm has high robustness to various types of attacks. Moreover, performance comparison prove the advantage of this scheme over some existing ones.The last proposed algorithm in the Chapter 5 is a blind robust adaptive video watermarking scheme using visual models in DWT domain. To improve and enhance the security, the watermark image is scrambled by Arnold transform. The scrambled watermark sequence is embedded by modifying coefficients of vertical and horizontal detail sub-bands of wavelet sub-blocks in host video. The visual model is developed by analyzing luminance masking, texture masking and spatial entropy masking. Experimental results show that the proposed scheme is efficient, has good transparency and robustness against various attacks such as filtering, noise, JPEG compression, and cropping and histogram equalization.