Research On Zero-Watermarking Algorithm With Robustness And Distinguishability

With the rapid development of network technology,people can browse and download a large number of digital information from the Internet in every day.As one of the important carriers of digital information,digital image plays an important role in information exchange and emotional expression.However,illegal people often use network software to maliciously tamper with and illegally copy digital images and spread them in a large number,which seriously destroys the copyright of digital information.The unpredictability of attack type and attack intensity in the network environment has a great impact on the stability of zero watermark.At the same time,the zero watermark between different images and similar images is difficult to distinguish effectively.Therefore,considering the robustness,distinguishability and information security,this paper improves the traditional singular value decomposition,differential entropy,visual cryptography and feature extraction methods,and proposes five robust and distinguishable zero watermarking algorithms to solve the problem of poor robustness and low distinguishability of zero watermarking in complex environment.The full text works as follows:Firstly,aiming at the problem that the features extracted from a single transform domain have poor stability under non-geometric attacks,a zero watermarking algorithm against non-geometric attacks based on Curvelet-DWT-SVD is proposed.By dividing the image into blocks and performing mixed domain transformation to extract more different frequency domain information,which can effectively improve the discrimination of zero watermark between different images,and reduce the impact of non-geometric attacks on the stability of carrier features.In order to further improve the performance of the algorithm against non-geometric attacks,the low-frequency domain of the sub-block is divided twice and singular value decomposition is carried out to generate the characteristic matrix,and XOR with the copyright watermark to generate a zero watermark.In terms of information security,Arnold scrambling is used to encrypt the copyright image,and XOR the encrypted copyright image with the feature matrix to generate zero watermark information,so as to improve the security of information.Secondly,aiming at the problem that the singular value of image changes greatly before and after geometric attack,a zero watermarking algorithm against geometric attack based on improved singular value decomposition is proposed.After the curvelet transform of the image,the low-frequency domain of the image is divided into blocks and singular value decomposition,and after singular value decomposition,the regulation factor is introduced to control the change amplitude of the maximum singular value.At the same time,by theoretical analysis and experiments verify,the robustness and distinguishability of algorithm is balanced by selecting appropriate regulatory factors.In order to solve the problem that the traditional encryption method is too single,a sub block mapping method is proposed to encrypt the copyright watermark,binary image 2 × 2 pixels mapped into an ascall value can reduce the size of zero watermark information by half and improve the security of zero watermark information.Thirdly,aiming at the imbalance between robustness and distinguishability of zero watermark under combined attack,a zero watermark algorithm against combined attack based on improved differential entropy is proposed.The variance ranking of all sub blocks is used to select feature sub blocks to reduce the impact of non-geometric attack and geometric attack on robustness.At the same time,the improved differential entropy is introduced into the calculation of sub block features to improve the distinguishability of zero watermark between similar images.In the process of copyright image encryption,the copyright image is processed in blocks,different encryption methods are adopted for different watermark sub blocks,and the key is distributed by Lagrange equation to achieve the purpose of copyright sharing.Then,aiming at the problems of poor robustness and single information of zero watermark under rotation attack,a distinguishable color zero watermark algorithm based on visual cryptography is proposed.By decomposing the color components of the color carrier image,each color component is constructed into a multi-dimensional tensor.At the same time,in order to improve the robustness of the algorithm,the expanded tensor is processed by block and discrete cosine transform to refine the carrier features,and the characteristic matrix is generated according to the symbols of the sub block discrete cosine coefficients.In the process of generating zero watermark,the principle of visual cryptography is used to generate public and private components,and combined with the carrier characteristics to generate color zero watermark information,so as to improve the distinguishability of zero watermark between similar images.Finally,aiming at the problem that it is difficult to distinguish zero watermark between similar images,a zero watermark algorithm based on regional XOR and ternary quantization is proposed.The difference of image features is expanded by using image frequency domain features and spatial features,so as to improve the distinguishability of zero watermark.At the same time,in order to improve the robustness of the algorithm,the image features and copyright information are redundantly constructed to correct the wrong bits of copyright information,when the watermark image is attacked.In the stage of copyright authentication,the third value is introduced to quantify the copyright information,so as to improve the visual effect of copyright information.