Digital Watermark To Optimize The Detector Design And Research

With the widely used and spread of digital medias on the internet, people can get onto the internet to conveniently spread, acquire and exchange information, copyright protection of digital media becomes more and more important. But the traditional security systems can only guarantee information transmitting process and can't satisfy other requirements of new conditions. As an effectively complementary method of traditional security, Digital watermarking is a new better resolution for copyright protection and media security.Digital watermarking is the technology which embeds copyright information into original multimedia and then depends on extracting watermarking or detecting watermarking to judge whether or not containing given watermarking, however extracting watermarking is very difficult with baleful attack for watermarked digital median and unknowing embedding algorithm, even if successfully extracting watermarking which also can't be identified. When facing this problem, we need watermarking detection to judge whether or not containing given watermarking which comes down to how to improve detection capability and reduce error probability. This paper does research on the algorithm design of watermarking detection about human visual system, detection theory and detection thresh, and then applies error correcting coding to the digital watermarking.The main research has tow parts:The first part does detailedly research on the digital watermarking detection.1,Based on analyzing and researching correlation detection and asymptotically optimal detection, using regression analysis and curve fit to improve asymptotically optimal detection which bases on the general Gauss distribution model. This paper devises a method to optimal the estimate of the shape parameter which can make this detection more validity and practicability and have more external detection data.2,Due to characteristic of the design of digital watermarking system, error probability always exist. The probabilities of false alarm and false rejection are affected evidently by detection thresh which is determined by idio-experience. This paper makes statistical analyze for the detection, then the detection thresh based on the characteristic of asymptotically optimal detector is theoretically deduced to gain a expressions of optimal detection thresh which make error possible of watermarking detection system lowest. This method has adaptive and objectivity, and improves detection capability.The second part does research on combine error correcting coding with digital watermarking.Because error correcting coding is usually used to improve robust watermarking for binary images, this paper applies Turbo code to watermarking algorithm and does research on the code-decode structure and capability. Using low error code rate and the characteristic, a digital watermarking algorithm which research the Turbo coding of grey value is designed based on Turbo code applied to grey image watermarking on DWT domain. It improves the robustness of algorithm. The different attack tests prove this method has good robustness of image on the different code rate of Turbo code.All the processed of the algorithms presented in this paper, including building, embedding, detecting, attacking, extracting and validating are realized by MATLAB7.0. The experimentations confirm security and usefulness of these algorithms.