Research On Robust Digital Image Watermarking Algorithm And Watermarking Protocol

With the rapid development of multimedia and network technology, multimedia information can be conveniently distributed, transmitted, and processed with help of computer. Besides its easy of our lives, the digital product has security and copyright problems. It is vulnerable to attack and forgery and hard to track the manipulations. Due to the copyright problem, digital watermarking technology is widely studied and concerned lately. The research and development of digital watermarking is used not only as proof of ownership, for copy control, transaction tracking, but also for the content integrity and recovery. As a multi-disciplinary technology, the study of digital watermarking is important academically, and also great potential in applications.The research of robust digital watermarking algorithm is still a hot topic, especially the geometrical resistant watermarking algorithm.To improve the accuracy of watermark extraction/detection and the robustness of the algorithim, focused on the three stages of a digital watermarking system,research of the generation, embedding and detection of the watermark is studied. Further, watermarking protocol is studied to promote the use of the watermarking algorithm.Noise is generated in the transmission, the attacks and signal processing steps of the watermark system. This may lead to the failure of detection and extraction of the watermark. To improve the effectiveness and accuracy of the system, the watermark is preprocessed in the generation stage and coded with error-correction approach. BCH code has strict algebra structure and strong error-correction ability, which approaches upper bound theoretically. The binary BCH code is implemented in this study. The encoding and decoding process have been optimized to produce high effective watermarks. The emulation has shown that the optimum approach has largely improves the effectiveness. Based on the optimized BCH encoding and decoding approach, a BCH coding watermark has been designed and implemented. It has been shown that the accuracy of watermark extraction is improved through redundant coding. The appraoch is useful to both meaningful and meaningless watermarks.It is a hot and hard problem current to achieve rotation, scaling and translation (RST) invariant watermarking. A robust watermarking based on wavelet modulus maxima (WTMM) approaches is developed in this study to meet this requirement. In this approach, the watermark is embedded to the WTMM of Mexican hat wavelet, which is invariant to rotation, scaling and translation. The watermark is detected with correlation analysis. The experimental results have shown that the algorithm is robust to image processing such as rotating, scaling, and shifting. The detectability of common attacks and JPEG 2000 attack is high.Based on the analysis of the current detection approaches, a new watermark detection approach is developed. It is based on the mutual information and robust to geometry attacks. It overcomes the point-to-point computation problem encountered in the correlation detection, which is tended to be changed after geometry attacks. On the other hand, the correlation detection is suitable to Gaussian noise detection and not effective to the non-Gaussian signal detection such as meaningful image watermark. Mutual information approach is adapted to handle this problem. The mutual information can describe the signal's higher order statistics and robust to geometry attacks. A watermarking scheme based on this algorithm is designed and implemented, and results have been demonstrated.In order to provide copyright protection in the practical transaction, digital watermarking protocol is studied. To resolve the collusion problem, the interaction mode and the watermarking generating method in Memon and Wong's protocol are revised. In the proposed protocol, the trading parties use the commutative encryption system to produce the watermark, and apply the homomophic encryption algorithm to embed the watermark and generate watermark image. As neither trading side has full watermark information, they can not collude with WCA. The homomophic encryption algorithm makes assure that the seller does not get the final watermark image and protects the benefits of the consumer. On the other hand, the buyer can not delete the watermark, which is helpful in tracking illegal copies and provides fairness.