Digital Watermarking Algorithm Based On The Theory Of Uncertainty

The digital watermark is to embed some digital information that can prove the copyright ownership or follow the tort to the multimedia data by certain methods.The information shouldn't affect the use of the original contents,isn't easily discovered by the human perceptual system again,only can be extracted by special detector.The watermark information can be the author's ID,the company marking,the text with special meaning and so on.The research on digital watermark algorithm could better realize the copyright protection of the digital product in network environment.Uncertain theory includes fuzzy sets theory,grey system theory and so on.It expresses and deals with the uncertainty of the knowledge in human cognition field.There are many uncertainties between embedding position and embedding strength in digital watermark,which uncertain theory could express well.This paper applies uncertain theory to digital watermark algorithm and does research on the algorithm design and encryption dealing about human visual system,fuzzy adaptive resonance theory and grey relational analysis.The main research contents are as follows:1.Digital watermark algorithm design(1)This paper makes a study of watermark's embedding position which is hot topic.Then it combines with human visual characteristic and presents a digital watermark algorithm which is adaptive in embedding position and embedding strength.The carrier image blocks are divided into four classes by brightness and texture.We could choose suitable middle frequency coefficients to embed watermark from every block which is based on the proportion between watermark and carrier image.At the same time the embedding strength is adaptively adjusted according to the classes.(2)At first the paper compares the digital watermark algorithm based on fuzzy c mean clustering with the traditional algorithm based on image classification and clarifies the former's superiority.But the slow convergence speed is a disadvantage for the algorithm based on fuzzy c mean clustering.To solve this problem,the paper presents an algorithm based on fuzzy adaptive resonance theory.This algorithm classifies the image suitably and executes quickly.The experiment proves that the anti-attack ability of the algorithm is better than the algorithm based on fuzzy c mean clustering.(3)The paper applies grey system theory to digital watermark and devises an algorithm based on grey relation and sampling theory.This algorithm enlarges the original image's edge and chooses nine lattices to be comparing sequences.The nine lattice is formed by every point as center.After this,the algorithm computes grey absolute relation degree,compares it to comparing threshold.The value of comparing threshold can be adjusted by the number of watermark information.At last,the suitable embedding positions for watermark are found.The experiment shows the algorithm has excellent performance.2.encryption dealing(1)The paper particularly analyses watermark design and preprocessing technology and realizes they should be enhanced in security.It presents a secondary chaos encryption algorithm for watermark image.Compared to traditional encryption process,this algorithm enhances the difficulty for attackers to illegally damage and the performance in secretness.(2)The paper analyses scrambling algorithms of digital images and particularly discusses their use, strongpoints and shortcomings.Then it presents an algorithm based on sampling theory.This algorithm is simple to use, the calculation quantity is small,the scrambling result is well.The original image could become disorderly without too many times to scramble.All the presented algorithms and encryption dealing in this paper are realized in Matlab.The experiments confirm security and usefulness of these algorithms.