A Study On Invisible Digital Image Watermarking Technology

This thesis is concerned with one of the most important branch of information hiding, the digital watermarking technology, with emphasis on its applications in digital images. Firstly, this paper introduces background of research of digital watermarking technologies, describes the basic concepts and classification of digital watermarking, as well as application fields and development of digital watermarking, formulates the design principles and methods of digital watermarking, and analyses the performance of digital image watermarking algorithm.Then, based on the theory of Discrete Fourier Transform, Discrete Cosine Transform and Discrete wavelet Transform, two invisible image watermarking schemes in transform domain are proposed, one is circular symmetric spread spectrum digital image watermarking based on chaotic sequences, the other is the hybrid image watermarking based on wavelet decomposition and discrete cosine transform. Finally, the good performance of the two algorithms is verified by some common image processing and attack experiments.Circular symmetric spread spectrum digital image watermarking using chaotic sequences as the watermark takes advantage of the Discrete Fourier Transform properties and spread-spectrum communication techniques to embed a circular symmetric watermark in the frequency domain. The watermark can be detected without the presence of the original image even when the watermarked image suffers geometrical attacks such as translation, rotation and scaling. Experimental results show that the proposed scheme has good transparency and highly good robustness to geometric transformations. In addition, robustness of the scheme against JPEG lossy compression, image enhancement, noise addition and statistical averaging attacks among others are provided.The watermarking scheme based on wavelet decomposition and discrete cosine transform using a binary pattern or gray-level pattern as the watermark, the scheme initially performs wavelet decomposition of the host image and the watermark with different hierarchy. After the pseudo-random permutation, each decomposed layer of the watermark is embedded into the DCT domain of corresponding decomposed layer of the host image. Extracting the watermark requires the original image, and its procedure is the reverse of embedding procedure. The existence of watermark is decided by observation visually and similarity measurement. Since watermark is added to visually more important regions of host image according to Human Visual System, and a pseudo-random permutation is performed in the wavelet domain before the watermark is embedded, which disperse the spatial relationship of pattern. Therefore, even if the shape or the major location of the watermark is estimated, it's extremely difficult to detect or destroy the watermark with the cipher of permutation unknown . Experimental results demonstrate the good robustness and transparency of the watermarking.