Algorithms Research On Domain-Based Image Watermarking Resistant To Geometric Attacks

The arrival of the era of the Internet, people’s life brought great convenience,all kinds of digital multimedia products born at this time, such as image、audioand video, and so on. Due to the attention of the people and the network open andsharing, that they become illegal people illegally gain, copy, modify, spread andprofit of object, seriously damage the legitimate users and the interests of thecopyright owner. In the face of such rampant illegal behavior, how to ensureeffective implementation in the network copyright protection and networkinformation safety, it is a current and important topic.Digital image watermarking has defects and shortcomings in the watermarkembedding capacity and resist geometric attacks robustness. According to theproblems, on the basis of transform domain algorithms at home and abroad, withdiscrete cosine transform domain, Contourlet transform domain and dual tree-complex wavelet transform domain as research scope and means, with improvethe ability to resist geometric attack watermark as purpose, this paper carried outthe research and improvement in the carrier image features extraction, watermarkpretreatment, embedding, extraction methods and properties, for the networkmultimedia works to provide the security of copyright protection.The specificworks are as follows.First of all, the paper gives an improved watermarking algorithm whichbased on the discrete cosine transform domain and double chaos system. Thealgorithm adopts Logistic mapping and Arnold mapping of combining thewatermark pretreatment method, emphasize the pretreatment process to completethe work of watermark security performance to strengthen. Because a good use ofthe characteristics of the chaos system, the image through of the discrete cosinetransform have more good security than the image processed after single chaossystem. In the second place, the paper proposed a double watermarking algorithmbased on Contourlet transform domain. From the theoretical basis to observe,transform later it got significant difference between the characteristics of sons.This is not the only with a whole past watermark embedding method, but weadopt different watermark embedding method, reach more ideal watermarkingsystem performance goals.Finally, the paper put forward a kind of digital image watermarkingalgorithm which combine hard C mean clustering and double tree complexwavelet transform. According to the human visual characteristics, only for thehuman eye sensitive dual tree-complex wavelet transform two low frequencysub-band to hard C mean clustering division, confirmed the embeddedinformation area. Then through the use of their local image correlation, adjust theembedding strength, change coefficient, and to realize the regional can embedwatermark embedding.Through the simulation experiment results show that theproposed algorithm has the feasibility and efficacy and safety.