Research On DWT-Based Adaptive Blind Watermarking Algorithm And Halftone Images Watermarking Algorithm

Along with the rapid development and broad applications of the multimediatechnology and network technology, it becomes an impendent problem to beresolved to protect the copyright of the multimedia digital products. The limitationof the raditional encryption technology which can resolve the copyright protectionproblem accelerates the appearance and development of the digital watermarkingtechnology. Digital watermarking technology is a new research field in the currentinformation safety technology field, and a kind of new technology to protect thecopyright, and to authenticate the source and the integrality of the works.Originators'information and individual symbol can be embedded into themultimedia data in the form of imperceptive watermark by the digitalwatermarking embedding system, while people can't see the watermark on theimage. Only by the special detector or computer software can we detect thehidden watermark.Wavelet translation not only matchs the characteristic of HVS (Human VisualSystem) preferably, but also is compatible with the JPEG2000 and MPEG4compression standard. Wateramark embedded with DWT has good visual effectand good robustness, so digital watermarking technology in wavelet domain is themain research direction in recent years. Halftone images often are used in book,magazine, newspapers, printer output and fax, so researches on how to embed theinvisible mark (such as corporation name, ownership information, production date,copyright protection and authentication information, and so on) to the halftoneimages have very important significance and practicality value. Although manyalgorithm have been proposed to embed watermark, most of them are aiming atgray image or color image, which aren't applied to halftone image directly. Andthe halftone images include only 0 and 1 two tones, so slightly alter may bring onthe image distortion, and the imperceptibility will not be satisfid. So the halftoneimage watermarking technology is the difficulty of the watermarking technology.After collecting and summing up a lot of literature about wavelet domain watermarkingtechnology and the halftone image watermarking technology, we carefully study and analyzemany image watermarking algorithms, and propose an improved adaptive blind watermarkingalgorithm to embedd readable watermark and a predigested halftone image watermarkingalgorithm.This paper consists of six chapters. The first three chapters mainly introducethe basic knowledge and the existing watermarking algorithm; auther's mainresearch works are introduced in the fourth chapter and fifth chapter and the lastchapter is the conclusion of the whole paper.The study meaning and the study background of digital watermarking areintroduced in chapter one and it is pointed out that digital watermarking is veryimportant to the copyright protection of the digital media. In this chapter, weintroduct the concept and the origin of the digital watermarking, and introduce thedevelopment status and the main application of the digital watermarking.The basic knowledge of digital watermarking and the mathematicsknowledge used in this paper are introduced in chapter two. This chapter describesthe digital watermarking embedding model, the detecting model, and thewatermarking system evaluating standard. Then we describe the waveletknowledge, such as the concept of wavelet, multiresolution analysis, Mallatalgorithm and 2D wavelet, etc.The classical digital watermarking algorithms are introducted in chapterthree. In this chapter, the watermarking algorithms are sorted two classes: spatialdomain watermarking algorithms and transformation domain watermarkingalgorithms. This chapter introduces the two-class algorithms carefully, andanalyzes the virtue and defect of these two-class algorithms. Finally, we points outthe existing problems in the watermarking field at present and the developmentdirection in the future.In Chapter Four, by analyzing the characteristic of detail subbandcoefficients of the image after DWT, we use the mean and variance of the detailsubbands to modify the wavelet coefficients adaptively to embed and extract thewatermark. We propose an adaptive blind watermarking algorithm basing on theDWT, and design gray and color two systems to realize the embedding andextracting watermarking. At last we make the simulation to this algorithm. Fromthe simulation results we make the conclusion that the algorithm proposed in thischapter may embed the binary image with lot information, and the watermark hasgood imperceptibility and robustness to many attacks, such as JPEG/JPEG2000lossy compression, additive noise, cropping, pixel-shift, and so on.In Chapter Five, a predigested digital watermarking algorithm for halftoneimage using stochastic error diffusion is proposed in this chapter. Using thismethod we can embed a black-and-white binary image into the halftone image.And we use two simple methods to extract the watermark from the halftone imagewith watermark. Simulation results show that the halftone image with watermarkhas good visual effect compared with the halftone image without watermark andwhen the halftone image with the watermark is severely stained or scratched, thedetection algorithm can detect the watermark efficiently. In Chapter Six, we summarize the whole paper, and show out some questionsto be resolved in the digital watermarking in the future. The main innovative work of this paper focus on the following two aspects: 1. Aiming at the defects of the existing watermarking algorithms in thewavelet domain, such as it is difficult to realize the blind watermarking extracting,and it isn't robust to many attacks, and the watermark can provide only a littleinformation, by analyzing the characteristic of detail subband coefficients of theimage after DWT, we use the mean and variance of the detail subbands to modifythe wavelet coefficients adaptively to embed the watermark. We propose anadaptive blind watermarking algorithm basing on the DWT, and design gray andcolor two systems to realize the embedding and extracting watermarking. Thewatermark in this paper is a meaningful binary image, and when we extractwatermark, the original image isn't needed. At the same time, the watermarkedimage is robust to many kinds of attacks, which is satisfied with need of practiceand has good practical value. 2. A predigested digital watermarking algorithm for halftone image usingstochastic error diffusion is proposed in this paper. Using this method we canembed a black-and-white binary image into the halftone image. And the sametime, we use two simple methods to extract the watermark from the halftoneimage with watermark. Simulation results show that the halftone image withwatermark has good visual effect compared with the halftone image withoutwatermark and when the halftone image with the watermark is severely stained orscratched, the detection algorithm can detect the watermark efficiently. But it hasa long distance to application. And this paper has some groping action to the three questions as follows: (1) How to resolve conflication between the imperceptibility and therobustness; (2) How to find more robust watermarking algorithm to resist the...