Research On Algorithm And Robustness Of Digital Image Watermark

Research on algorithm and robustness of digital image watermarkThe digital watermark is the information such as pictures, characters and numbers thatare embedded in the digital products. But I consider it as a process in which theverification information is embedded or extracted under the condition that we do notchange the property and application of the product. Since1990, digital watermarkinghas been a research hotspot in information security and information hiding in ourcountry. It’s widely used in terms of copyright protection, authentication,confidentiality, communication, etc. In the period of learning in Jilin University, Iachieved two types of watermark algorithms which are the discrete cosine transformdomain (DCT) and singular value decomposition (SVD), and studied their robustness.Compressed sensing is a new theoretical basis in the field of signal processingresearch. I use this theory in digital watermarking field, and study it’s robustness, andtest the robustness margin.In this paper, I first introduce meaning and research background of the topics. Aftertalking about the basic conceptions of digital watermarking systemically, I mainlynarrative the process of the learning on digital watermarking and completing theinnovative. I write this paper in the order of learning process of myself:Watermark algorithm in DCT domain: I complete three kinds of additivewatermarking algorithm in both whole DCT and block DCT. In the whole DCTdomain, I complete a simple additive watermark process; in block DCT domain, Icomplete the watermark algorithm after ordering in middle-frequency, andself-adaptive algorithm based on human visual model in low-frequency. Theexperiments show that: the whole DCT domain watermark algorithm can embed alarge amount of information, and meanwhile the algorithm can maintain certainrobustness. The robustness of algorithm in block DCT domain is stronger than that inwhole DCT. I use the self-adaptive algorithm in the low-frequency coefficients ofto ensure the robustness, at the same time, a better transparency. During the analysisfor the robustness of the watermark, and the robustness margin can be used intuitively to express the robustness which the watermarking algorithm may reach and to meetthe adaptability requirement of the practical application;SVD algorithm: I complete the embedding and extraction of watermark in additiveimages by doing singular value decomposition of them. The experimental resultsshow that: digital watermarking algorithm based on singular value decomposition hasa strong robustness, and with high degree of transparency, so it can be applied to avery wide scope.Compressed sensing algorithm: Combining with the compressed sensing theory andwatermark, this paper presents a algorithm in which the watermark operation iscomplete in the observation domain of the sparse decomposition of the images. Theexperiments show that the algorithm has a good transparency, a strong robustness, agood security protection capability of the embedded watermark, which indicate thefeasibility of this algorithm.Because different practical applications of digital watermarking have differentrequirements, no one algorithm can meet all. There are verities of algorithms that canmeet some requirements in each way. In this paper, I study some watermarkingalgorithms and their robustness. In terms of the robustness, geometrical attach is themain difficulty which will be the focus of my follow-up study.