| With the rapid development of computer network,digital image is widely used,but at the same time,the problem of copyright protection and illegal transmission of digital image is becoming more and more serious.Digital image watermarking technology is one of the effective means of digital image copyright protection.In recent years,the research of digital image watermarking technology has greatly promoted the application of watermarking technology.However,for the time being,a watermarking algorithm does not necessarily apply to all images,that is,for some images,the algorithm can show good performance,but for some other images it does not necessarily show the same good performance.So,for a given carrier,how to determine whether the algorithm is applicable to the carrier,the existing literature rarely concerned about this issue,which is a difficult area of watermark research.Therefore,this paper mainly aims at the application of carrier evaluation for the watermarking algorithm,hoping to provide some useful help for the practical application of watermarking.The main work of this paper includes three aspects:1.The change of the maximum singular value of the spatial image and the the JPEG domain image are analyzed.By the singular value perturbation and K_S hypothesis test,it can be concluded that the maximum singular value of the aerial image is affected by Gaussian noise,and its maximum singular value does not exceed the maximum singular value of the additive noise matrix and follows the Gaussian distribution.In addition,according to the image DCT coefficients obey the Laplacian distribution,and after the linear transformation of its distribution law unchanged,and JPEG compression is a linear change process on the DCT coefficient,can be launched after the JPEG image quantization DCT coefficient obeys Laplacian distributed.And the Laplace distribution is a special Gaussian distribution,and the distribution of the same distribution after the distribution of the same.Therefore,after the JPEG image overlaying the Gaussian noise attack or after the JPEG re-compression attack,the change of the maximum value of the DCT coefficients after the quantization is obeying the Laplacian distribution.2.Aiming at the problem that the carrier can be evaluated by the airspace image watermarking algorithm,an improved robust watermarking algorithm is presented in combination with the variation of the maximum singular value of the image after the image noise.In this approach,the watermark component is added to the largest singular value of the corresponding image block through dither modulation.The robustness of this method towards attacks and the watermark invisibility can be theoretically evaluated.The ability of the watermark to resist noise is determined when the carrier image meets the condition of the statistical hypothesis test.After deriving the peak signal-to-noise ratio,it can be shown that watermark invisibility is dependent on the size of the image blocks and the dither quantified factor.Extensive experiments are performed to further confirm the theoretical analysis and to demonstrate the ability of the proposed algorithm to robustly resist noise and JPEG compression attacks.3.A scalable JPEG image robustness watermarking algorithm is proposed to solve the problem that the carrier image can be evaluated by the JPEG domain image watermarking algorithm.Combined with the probability distribution of DCT coefficients after JPEG image quantization.Firstly,the corresponding quantized DCT coefficients are selected according to the embedded secret key and do singular value decomposition for each quantized DCT coefficient blocks.Then,the watermark component is embedded into the maximum singular value of the corresponding quantized DCT coefficient block by dither modulation.In this paper,the relationship between watermark invisibility,embedding step and original image quality factor is given.Based on the assumption that the quantized DCT coefficients obey the Laplacian distribution from the perspective of watermark evaluation,and the change of maximum singular value after JPEG compression and Gaussian noise attack is obeyed the Laplacian distribution,and theoretically gives the ability of watermark resistance attack.The experimental results show that the algorithm is correct in the theoretical analysis of the peak signal to noise ratio,the robust of Gaussian noise attack and the JPEG re-compression attack.The experimental results show that the algorithm is invisible,robust and so on Aspects of the obvious advantages.At the end of the thesis,the paper summarizes the whole paper and points out some problems to be further studied. |
