Digital Image Watermark Algorithm In Frequency Domain Based On Chaotic

The digital media has become a main way for information on communication along with the wide use of network and has played an important role in people's life. As side effects of the popularity of digital representation and distribution over network, some individuals or groups copied, transfered, amended and even sold the digital media without the permission of copyright owners, which greatly damaged the copyrights and commercial benefits of owners. Therefore, the problem how protect the copyright has become a hot one in research work. In the dissertation, we mainly study the technologies of digital watermarking in frequency domain.The encryption of digital image was derived from the early classical encryption theories. The purpose of it was to change a given image into an out-of-order image in space domain or in frequency domain by a rule so that the real information of the image itself was hidden. Digital watermarking technology as an effective means of knowledge copyright protection is widely studied and used. It embeds some secret messages—watermarks into the media to protect the property or integrality of the working so that digital produces can resist being copied, tampered and so on.In the paper, the technology and application of the digital image watermarking are deeply analyzed and studied, the main work is as follows:(1) We make a recapitulative summarize and analysis for the basic knowledge of digital watermarking technology from its origin, basic character, sort, common model, main application field and so on.(2) In the paper, two new methods on digital watermarking are finished: the first is based on discrete wavelet transform and the second is based on discrete cosine transform. In order to enhance the watermarking encryption and vision effect of the extracted watermarking, firstly a two-value watermarking signal should be dealt with by shuffle, the times may be as a part of the key, and then is modulated by chaotic sequence. In the first one, watermarking is embedded into the chose sub-image from original image. Many DWT watermarking algorithms based on HVS embed watermark into the intermediate frequency band of DWT, but we embed watermark into the high frequency band in order to enhance the embedment intensity and hold the watermark'simperceptibility. In the second one, watermark is embedded into the low frequency coefficients of DCT, because the low frequency part of the image collects a majority of energy of the original image and has a biggish value. Attacker will inevitably affect the quality of the image when he destroys watermarks. We can gain enough robustness and imperceptibility by this way. In this way, we realize blind extracting algorithm.(3) In order to further evaluate performance of the algorithm, simulation experiment is done include the attack of normal signal processing and geometric operations, such as common noise, crop, JPEG compress, image brighten and so on.