Research On Digital Watermarking Algorithms For Images

With the coming of the information era, Internet heat sweeps across the whole world. At the same time, E-commerce and E-government affair develop very fast. Thus, Information security has becoming an eye-focus problem. Traditional encryption methods can solve this problem to some extend, but with the fast development of the powerful computing technique, it becomes easier and easier to decipher. It is no longer the only method to protect the secrete files through lengthening the key. As a new information security technology, digital watermarking technique embeds watermarks into the cover media to protect the copyright or ownership of the digital commodities. It can be also used to authenticate digital commodities. This dissertation focuses on research on watermarking algorithms of digital images.This dissertation consists of eight chapters. Chapter 1 and 2 are the fundamentals. Chapert 1 includes the purpose of the research, the history of the watermarking technique, the development status of multimedia watermarking and its application domain. At last, the main research works in this dissertation are presented. Chapter 2 surveys systematically the characteristics and classification of watermarks, the basic frame of the watermarking system and attacks classification and according protect strategies. From chapter 3 to chapter 7, some watermarking algorithms are described and analyzed. Each chapter includes one algorithm. The main contents are:(1) Earlier algorithms were generally in the spatial domain because the computation is simple and large amount of information can be hidden in the cover image. In this dissertation, a blind watermarking algorithm in spatial domain based on modulus is presented. The modulus operation, which is an invertible operation, decides the embedding position, thus illicit user cannot extract the watermarks without the private key.(2) Because geometric transformation may destroy the synchronization of the embedding and detecting processes of watermarks, it is a difficulty how to resist the geometric attacks in watermarking technique. In this dissertation, a blind watermarking scheme that is robust to geometric attacks is presented. The original image is normalized according to the geometrical moments, then a parameter in the normalized image computed. At the same time, the same parameter in some normalized after StirMark-attacked images are also computed. It is important to design the watermark elaborately to ensure the parameter change in a pre-defined extend. For an image, if the parameter in its normalized version is in this extend, then we consider that there is watermark in it; or else, there is no watermark. This scheme is very simple and without resorting to the original image when detecting the watermark.(3) In a CDMA communication system, the Walsh code is used to divide different users. In this dissertation, a spread spectrum watermarking system using orthogonal Walsh code is studied. Different from the CDMA system, the same pseudo random sequence is modulated by three orthogonal Walsh codes. Then the compound signal is embedded in the DCT domain of the original image. Experiments show that this system has high performance in resist attacks.(4) Taking into account the features of human visual system, a watermarking scheme in DWT domain is designed to make a compromise between the invisibility and the robustness of watermarks. A binary image watermark is embedded into a set of important coefficients that are chosen through zero-tree coding algorithm. And the embedding strength is no more than the just noticed difference (JND). Consequently, watermark is hidden in the cover image with enough strength while without resulting in degradation of the original image.(5) Gray-scale image is better than binary image for differentiation, so a watermarking algorithm based on integer wavelet lifting scheme is proposed, which embeds gray-scale image into cover image. Firstly, the gray-scale image is messed and decomposed into several bit planes. Then the cover image is blocked and transformed by integer wavelet lifting scheme. Thirdly, those bit planes are embedded into transformed coefficients. Experiments show that this scheme is robust to general attacks, especially to JPEG and shearing attacks.In Chapter 8 the whole dissertation is summarized and possible study contents in the future in watermarking technique are also suggested.