Encrypted Domain Reversible Image Watermarking Algorithm Research

In this thesis,we focus on the research of reversible watermarking algorithm with low distortion and large capacity.Two reversible watermarking algorithms in encrypted domain are proposed.The main contributions in the thesis are as follows:An adaptive reversible image watermarking algorithm based on histogram shifting and difference expansion in encrypted domain is proposed.It is a separable reversible watermarking algorithm.In the traditional approach based on the difference expansion and histogram shifting,two adjacent points form the expansion center.While in practical application,the size of a host image is quite larger than the watermark image.If the watermark image is embedded by the traditional method,especially when the payload is smaller,the distortion is larger.Therefore,on the basis of the traditional method,the method of adaptive expansion center is proposed in this thesis.That is to say,in the process of watermark embedding,the extended boundary is kept unchanged,and the extension center is selected according to the size of the embedded watermark.In a certain payload,the amount of error correction will be reduced,which can effectively improve the quality of the watermarked image.The experimental results show,comparing with the traditional histogram modification method,the proposed method can improve the peak signal to noise ratio of the watermarked image.The distortion of the watermarked image is reduced.A reversible image watermarking optimal algorithm based on multi-histogram in encrypted domain is proposed.The purposes are to improve the embedding capacity and the quality of the watermarked image.The basic ideas are as follows: at first,pixel errors are divided into some groups according to their complexity.And histograms of error groups are constructed.Then,by analyzing the cause of the distortion in the image,the model of the relationship between the image distortion and the embedded payload is constructed.The embedding position parameters of each histogram are obtained by solving the model.Finally,the watermark is embedded in multiple histograms according to the obtained parameters.By using the grouping method,the experimental results show that the embedding positions are well hidden.So the anti-attack ability is improved.Combining optimal algorithm,the modified number of errors is also reduced.Thus,the embedding distortion can be minimized and the quality of the watermarked image can be improved.The sum of the peak points is also larger than the peak points of the single histogram,and the embedding capacity is also improved.At the same time,by using the optimal algorithm,the higher complexity histograms will be first chosen to embed the watermark.That is,the texture and boundary region of the image is first chosen to embed watermark.So the invisibility of watermark can be ensured.In a word,the reversible watermarking technology of digital images is studied,and two reversible watermarking algorithms based on histogram in encrypted domain are proposed in this thesis.By using above two algorithms,the watermarked image distortion can be effectively reduced in the case of a certain payload.The purpose of improving the quality of watermarked image and the capacity are also achieved.