Research Of Frequency Domain Digital Watermarking Based On Perceptual Characteristics

With the popularity of Internet technology, information technology not only provide agreat convenience of access to multimedia information, but also make the informationexchange between the breadth and depth of never had. Information Hiding as the feature ofcamouflage emerged as the times require in information security technology has become aneffective method for covert communication and copyright protection, and has beenextensively studied and applied internationally. Application of information hiding ininformation security system mainly concentrate in digital copyright protection, pirate tracking,identification in integrity and authenticity, and recovery etc.. As an integral part ofinformation hiding in the field of multimedia security technology, digital watermarking is theeffective method of realization of digital copyright protecting and tracking tort. As well as itsinherent characteristics, such as secret communications, implicit tagging and identifyingauthenticity, has become an important research in information security field.Based on the analysis of existing digital watermarking models and focus on thewatermark algorithm, image adaptive watermark embedding in embedding capacity,robustness and invisibility and so on, combined the human visual system and grayscaleimages in a number of characteristics in the frequency domain is studied. And the systemmodel of digital image watermarking is reach focus. The model increased the watermarkingcapacity, improved the system performance, enhanced the anti-attack capability and achievedthe blind watermark extraction. The main contents of this paper are as follows:1. The robustness of watermarking which embedded in spatial domain is poor, and easyto damage or to crack. Aimed at the problems, a digital watermarking model based onfrequency domain which has strong robustness and good invisibility is proposed in this paper. Analyzing performance prediction of watermarking system, simulation and experimentalresults show the effectiveness and feasibility of the model, so it is suitable for the imagewatermark embedding.2. Using chaotic mapping or Zigzag scanning technology to scramble and dimensiontransform the original watermarking information, the watermarking is encrypted through aspecial equation to enhance the security of watermarking information. By the method, even ifthe image is detected watermarking by destroyer which can not be extracted from the real.3. Traditional digital watermarking model is considered the merits of the image qualityof embedded without the visual angle. In-depth analysis of the watermarking image structureand the basic principles of digital images of human visual model, the combination of both iscalculated to obtain the threshold to achieve the purpose of embedding, and improve thewatermark invisibility ultimately.4. Frequency-domain transform coefficients of an image distribute according to thefrequency. Based on the comparison of past literature reference, aimed at different frequencydomain transform and threshold in content3, self-adaption selected different frequency bandsand the different coefficients for digital watermark embedding positions. The embeddingcapacity has been increased.5. Using the quantization of parity characteristics of the threshold for special treatment,the watermark embedding algorithm is proposed and the ultimate digital watermarkingmodels are established. The model does not require the original image and watermarkingimage in the watermark extraction, while it achieves the blind watermarking extraction andimproves the experimental operation rate.