Resynchronization And Fountain Code Based Robust Image Watermarking Research

Digital watermarking,as an effective means of copyright protection,has developed more than 20 years,but not yet mature.There are many problems to be studied and discussed in theory and practical application,such as watermarking evaluation criteria,geometric attacks,embedded capacity,blind extraction and so on.The geometric attack is considered to be the bottleneck for the application of watermarking technology,how to effectively resist geometric attacks on the basis of keeping high capacity,which is still a challenging task.At present,most watermarking algorithms are robust to simple geometric attack,but is poor for the cropping and the combined geometric attack,and has small capacity.To this end,an algorithm with higher capability in resisting the cropping and combined geometric attacks,as well as larger embedding capacity is proposed.The main achievements of this paper are summarized as follows:1.In view of the problem that most robust watermarking algorithm is poor for the combined attacks of rotation,scaling,translation and general affine transformation,the characteristics between them are analyzed in this paper,and the relationship and generality between them are found,namely,these attacks are the global geometric and we can use affine matrix to unify the characteristics between them.Therefore,the affine matrix is estimated accurately in the proposed scheme.Firstly,feature points from the attacked image are extracted with SURF algorithm,then they are matched with a small amount of feature points of strong robustness,which are extracted from the original image.The matching point pairs are further selected by nearest neighbor/sub-nearest neighbor ratio and RANSAC algorithm.2.In order to cope with the pixel shifting and interpolation error in the process of geometric attacks and restore and improve the embedding capacity,this paper using the method of quantization of parity in the image space domain embedding a watermark,precise positioning of the watermark in the transform domain,while the fine tuning RST parameters,reduce the bit error rate of watermark embedding.Increased capacity.The proposed scheme embeds a layer of watermarking in the spitial domain,to locate the copyright watermarking accurately in the transform domain,in addition,reduce the bit error rate and increase the embedding capacity.3.In view of the similarity between the digital watermarking system and the communication system,and most watermarking algorithms are fragile to cropping attack,the watermarking information is encoded by LT-code,which makes full use of the erasuring performance of LT-code to resist cropping.Finally,the experimental results show that the proposed algorithm not only has large embedding capacity and good transparency,but also can resist many kinds of attacks.