Construction Of Non-separable Wavelets And Its Applications In Digital Image Watermarking

Nowadays,with the rapid development of technology,the digitization of multimedia products and the universalization of the network have provided great convenience for accessing information(such as audio,pictures,videos,etc.),improving the efficiency of digital information transmission and accuracy during transmission.But it is also because of the convenience and speed of the transmission method of multimedia digital information that the problems of copyright protection and information security have arisen.Due to the lack of corresponding protection mechanisms,shared multimedia information is particularly vulnerable to illegal copying,tampering,and forwarding,which can create a gap for piracy and infringement by some illegal individuals or organizations,and seriously damage the legitimate interests of the copyright owner,therefore,digital image watermarking technology is a hotspot for in-depth exploration in various countries.In the field of digital image watermarking,researchers have proposed many digital image watermarking algorithms,such as the least significant bit(LSB)method,discrete cosine transform(DCT)method,tensor product wavelet transform(DWT)method,etc.Quantitative product wavelets can't capture singular points in all directions,which has become an obstacle to further improving the robustness of the watermark.Non-separable wavelets are isotropic,which makes up for the lack of tensor product wavelets in all directions.As far as we know,no one has applied two-dimensional two-channel non-separable wavelets to the embedding of image watermarks.Existing people have proposed image watermarking algorithms based on four-channel non-separable wavelets.The watermark is embedded in high-frequency sub-pictures,and the watermark is less robust.Therefore,the main research work of this paper is as follows:This paper focuses on two-dimensional non-separable wavelet transforms.By analyzing the construction method of non-separable wavelet filter bank,a two-dimensional two-channel diagonal band filter construction method is proposed,and two-channel and four-channel non-separable wavelet filter banks are constructed.Two digital image watermarking schemes using non-separable wavelets are proposed.The main work is as follows:(1)A watermarking algorithm based on two-channel non-downsampling nonseparable wavelet is proposed.This algorithm performs arnold transform on the watermark image,contourlet transform on the host image,takes its low frequency to perform two-channel non-downsampling non-separable transform,and then takes its low frequency block to perform singular value decomposition(SVD),after singular value matrix maximum is compressed by u-law,the watermark is embedded by QIM.This algorithm solves the problem that the original image can't be obtained,and the robustness is also improved,especially for hybrid attacks.(2)A watermarking algorithm based on four-channel non-separable wavelet is proposed.The algorithm first performs arnold transform on the watermarked image,then performs a multi-scale analysis on the host image using fourchannel non-separable wavelets,and performs SVD on its low-frequency subimage and scrambled transform watermark image,finally,the watermark is embedded on the singular value matrix of both with the principle of addition.The isotropy of non-separable wavelet is used to enhance the resistance of the watermark to attacks,especially the resistance to rotation attacks.By comparing the two methods proposed in this paper with the tensor product wavelet transform-based method,the DCT-based method and the existing four-channel non-separable wavelet transform method,it can be seen from the experimental results: The two methods proposed in this paper have better robustness.