Research Of Authentication Technology Image Content Based On Semi-Fragile Watermarking

In the era of rapid development of information technology,various means of information dissemination,image processing technology changing with each passing day,multimedia data is convenient for storage,sharing and transmission,which has enriched the ways for people to obtain information.But at the same time,because multimedia data is easy to be copied and modified,and its data volume is large and with high redundancy,it is difficult to track the modification of the characteristics of the digital media.Security has become a major problem in the information process,which will seriously hinder the development of the information industry.Therefore,how to protect and detect the authenticity,integrity and reliability of image content in the context of complex information blending becomes a problem that we urgently need to solve.In this thesis,discrete wavelet transform was used to transform image signal into wavelet domain,combined with the influence of human vision system on image vision,a semi-fragile watermarking algorithm that can realize image content authentication and tamper localization was proposed.Firstly,the source of watermark information to be embedded was determined,and meaningful or meaningless watermark information was selected according to the demand.In this dissertation,the image authentication function was required.The selection of watermark information should be related to the image content and include as many image features as possible,so as to identify whether the carrier image was tampered.In order to further improve the security of the algorithm,the watermark information was encrypted by Logistic chaos to generate the secret information to be embedded.Secondly,considering the energy distribution characteristics of the image after discrete wavelet transform,the secret watermark information was embedded into the horizontal and vertical detail components of the two-layer wavelet transform.In order to make the distribution of watermark signal more uniform and reasonable,the pseudo-random number generator was used to determine the embedding position of watermark information.The quantization degree difference of watermark information embedding was determined by combining image block itself and the embedding intensity and quantization step of watermark information were determined by combining critical noise threshold.Under the guidance of human vision,the energy of digital watermark signal was reasonably distributed,which improves the invisibility and compressibility of watermark information.Finally,the embedded watermark information was extracted from the wavelet coefficients and compared with the image watermark information after decryption.In order to improve the accuracy of tamper location as far as possible,sparse points were removed from the tamper matrix,small noise points were filtered out,and the tamper location was determined by combining the corresponding relationship between the tamper matrix and the image feature information.In order to verify the practicability of this algorithm,the invisibility,robustness and similarity of images were evaluated respectively,and the experimental results of JPEG compression attack,noise attack and shear attack on images were analyzed.The algorithm proposed in this dissertation balances the contradiction between the robustness and vulnerability of semi-fragile water and the contradiction between the information content of watermark embedding and the anti-attack performance.To some extent,this algorithm overcomes the limitations of vulnerability watermarking and robustness watermarking,and does not need to provide original image when verifying the authenticity of image content,thus reducing the dependence of watermarking detection on original carrier data.Experimental results show that the proposed algorithm was robust to compression attacks and noise attacks,and fragile to shearing attacks.In this dissertation,the critical noise threshold was taken as the influence factor in the embedding strength of watermark information and the random embedding method of watermark information,which improves the transparency of watermark information.Compared with the same type of algorithm,the signal-to-noise ratio of image was improved by about lOdB.Enhanced anti-jpeg compression performance,especially for the image in the case of high compression rate,tamper rate reduced by 5 to 10 percentage points,to achieve the image tamper location.