Research On Key Techniques Of Image Reversible Data Hiding Error Prediction

With the popularity of the Internet,the dissemination and use of multimedia information is becoming more and more widespread.At the same time,the security of digital information has become increasingly prominent.Information hiding is a technical means to effectively solve the problem of digital information security.It hides secret information in the public carrier to achieve the purpose of protecting information.As an important branch of information hiding technology,reversible information hiding technology can not only extract the embedded secret information completely,but also ensure that the carrier signal can be restored nondestructively.Reversible information hiding technology is applied to more sensitive scenarios,such as military,medical,legal forensics and other areas that do not allow any errors.In this paper,the key technology of image reversible information hiding error prediction is studied,and the problem of enhancing the embedding capacity of reversible information hiding technology is discussed.The main research work and innovative achievements of this paper can be summarized as follows:(1)A reversible information hiding error prediction algorithm based on multiple linear regression is proposed.Unlike most prediction algorithms,which only use the simple arithmetic combination of the pixels around the target pixel to realize the prediction of the target pixel,this algorithm uses the consistency relationship among the pixels in the local area of natural image to adaptively learn the intrinsic relationship among the pixels around the target pixel,and establishes a multiple linear regression prediction function satisfying the close correlation of the surrounding pixels.Matrix,and use the optimized linear regression function to achieve the target pixel prediction,improve the accuracy of the prediction algorithm.(2)A constrained multivariate linear regression reversible information hiding error prediction algorithm is proposed.Based on the multiple linear regression algorithm,the target pixel value can be predicted accurately,the error value can be reduced and the prediction accuracy can be improved.However,in areas where the image texture is more complex or the pixel value varies greatly,the predicted value will still have large deviation and produce a large number of singular values.In order to reduce the singular values generated by the multivariate linear regression algorithm and improve the prediction accuracy,a constrained multivariate linear regression prediction algorithm is proposed.(3)A reversible information hiding error prediction algorithm based on multivariate logarithmic regression is proposed.Firstly,the image is divided into "point set" and "cross set" by diamond algorithm for two-stage embedding,and then the pixel value is predicted by non-linear multivariate logarithmic regression.