Research Of Image Watermarking Based On Hypercomplex Fourier Transform And Visual Perception

With the continuous progress of the process of information, the further researches in many fields such as computer vision, digital communication and multimedia technology make information extraction, analysis, creation and the form of dissemination more convenient. Due to many advantages of easy to product, modify, storage and disseminate, so digital media works are more and more widely used ; At the same time, digital multimedia information is prone to be copied and altered illegally. Therefore, how to protect the copyright and integrity of digital multimedia in the modern information environment has become a serious problem. Image digital watermarking technology, as an effective means of image authentication and copyright protection, has been paid more and more focus.In this paper, according to the typical copyright protection issues that news, advertising, graphic arts and color content of digital works are frequently infringed, a method of color image digital watermarking based on visual significant under the overall framework of the ultra-complex frequency domain is proposed, in this paper, we study how to extract visual significant region of the color image and how to use correlation between ultra-complex frequency domain coefficients in these areas protected on behalf of digital work content, and then we test watermark embedding, extraction and anti-ability on the robustness of the copyright ,The main work is as follows:First of all, on the basis of visual areas that is often a significant areas to be protected copyright works, combined with significant detection methods of ultra-complex frequency domain, an automatic extraction algorithm that multi-scale visual significant using phase spectrum in super complex frequency domain is proposed, which constructs color hyper-complex quaternion in HSV color space that is more consistent with human visual perception, and using human visual's multi-scale resolution performance and frequency phase spectrum to reflect the image structure information and other characteristics, and then achieving rapid detection algorithm of significant region in a color image. The results show that: In most cases, the proposed algorithm is superior to the traditional saliency extraction algorithms, and accurately detects significant "content" area in an image.Secondly, In the ultra-complex Fourier transform domain of the color vector space, we study the digital watermarking algorithm in case of users customized copyright protected areas and automatic extraction protected area.After obtaining the user interaction specified copyright protected areas or automatic extraction significant regional, firstly the algorithm performs Fourier transform of the obtained region and achieve chaotic scrambling of copyright watermarks; then according to the watermark scrambled, frequency domain coefficients of frequency band are quantized in the region, the copyright watermark embedding is implemented. The anti-process is the watermark extraction process. Because of the almost consistent automatic extraction significant region with user-specified content area in most cases, the algorithm can automatically embed and extract copyright watermark, which is very practical.Finally, experimental results show that the algorithm has good performance of the hidden and the resistance to attack, which has a very good protection for copyright and integrity of color image content and a strong practical value.