A New Watermarking Algorithm Based On Discrete Wavelettransform And Singular Value Decomposition

The information exchange based on computer networks has provided great convenience to the usage and spreading of the digital multimedia products. However, the digital products are illegally copied and tampered very easily, which makes its information security and copyright protection a factual problem to be solved urgently. Taking use of the redundant data and randomicity of the digital products, the digital watermarking technology embeds the unperceivable but recognizable secret information-watermark containing the copyright information into the digital products, which becomes an effective method to solve the digital information security problem.In the thesis, the basic principle and the research status worldwide of the digital watermarking technology are introduced briefly. And the principles of discrete wavelet transform and singular value decomposition (SVD) are presented, on which a novel digital watermarking algorithm based would be approached in the thesis. Firstly, relavant coefficients which lies in the subband of grey image after discret wavelet transformation would be choiced to compose a matrix according to human visual model. Secondly, singular value would be took out from the matrix by the method of singular value decomposition (SVD). Finally, the watermark information which would be generated with sigular value and random sequence would be embeded in the matrix. It would be benfit for the invisual performance of algorithm that we embeded watermark information in coefficients which have been choiced by human visual model.The experience of test show the algorithm perform well in it's invision propery an have quit high capcity in anti-compact attack. And also, the algorithm is sensitive to other kinds of hostile attacks,such as median filter attack. To sum up, It's a really good semi-fragility watermarking arithmetic.