Research On Digital Watermarking Algorithm Based On Video

The rapid development of the network provides the conditions for the generation,storage and dissemination of digital products,multimedia technology also provides us with a wealth of resources.However,in such a superior environment,but also produce intellectual property and other issues.Copy,transmission and other convenient conditions of transmission makes the phenomenon of illegal distribution rampant,some criminals will use the network to piracy of digital works,tampering,and sale to other people to seek benefits,resulting in copyright owners of economic losses and rights violations.For maintaining the rights and interests of copyright owners,digital watermarking technology is widely used in copyright protection,copy control,fingerprint identification,content authentication,broadcast monitoring and other fields.Digital watermarking technology is to embed the watermark information into the digital file to be protected.If the digital product is not used normally,it can prevent the content from being copied and changed by encryption and other methods.As the scope of the spread of the network expand continually,if the current digital works of piracy,tampering and other issues to be taken seriously,then the consequences will not be imagined,therefore,digital watermarking technology has become more concerned about the current research direction.And the current film and television industry piracy problem is particularly serious,digital watermark research based on the video is the focus of watermark research.On the one hand,it is a problem that the robustness may restrict the invisibility of watermarking.Generally,the watermark embedding information is enhanced and the robustness of the watermark is enhanced,and the quality of the carrier signal is decreased and the invisibility is deteriorated,and vice versa.On the other hand,the watermark algorithm is not strong enough to resist to a variety of attacks,especially geometric attacks: robustness is an important aspect of watermark evaluation,how can the watermark in the case of various attacks by the outside can still be extracted effectively is a issue need to focus on.At present,there are still many problems of digital watermarking research based on video.In this paper,some problems of video watermarking are discussed and studied.The contents of this paper mainly include the following aspects:(1)For the problem of the difficult to balance the robustness and invisibility of video watermarking technology,proposes a zero-watermarking algorithm.The algorithm uses zero-watermark scheme to solve the problem of performance balance.For the problem of the whole process of dealing with the large amount of video data with high time complexity and the frame selection characteristic is not enough.The key frames of the video sequence are extracted by Euclidean distance method,and the key features of the video key frames are obtained by the idea of moving target detection.For the problem of the computing complex of3-D discrete cosine forward and reverse transform(3D-DCT),the zero-watermark feature sequence is constructed by using the coefficients in pseudo-3D-DCT transform domain.The feature sequence of zero-watermark not only has strong uniqueness to ensure the safety of the watermark,but also the construction of the zero-watermark does not need to be inverse3D-DCT.The computational complexity is lower than that of the normal 3D-DCT and pseudo3D-DCT watermarking algorithms.The robustness and invisibility contrast experiments are carried out under different attacks to verify the effectiveness of the algorithm.(2)Aiming at the problem that the video watermarking algorithm is not strong against geometric attacks,a video watermarking algorithm based on SIFT operator is proposed.Since the performance of the SIFT descriptor is better in many cases,the SIFT descriptor is selected for the feature point correction after geometric attack.Comparing with the Wavelet Transform,Contourlet Transform has some advantages and has better performance in watermarking algorithm.Therefore,the key frames are decomposed by Contourlet transform,the parts of low frequency are grouped in time direction and decomposed by one-dimension DCT,and the watermark is embedded in each group of transformed coefficients.When the watermark is extracted,the final watermark is determined by the sum of the values at the same pixel position of the watermarks extracted by each group to improve the correctness of the watermark.The experiments are done on the algorithm embedding and extracting process,and the robustness and invisibility of different algorithms are compared.It is verified that the algorithm has some progress for the previous algorithm.