Research On Applications Of Compressive Sensing Theory To Image Protection

Compressive Sensing(CS),as a newly proposed signal acquiring theory,has drawn much attention in many fields.Interestingly,CS framework can be used as a cryptosystem.In contrary to the traditional cryptography,CS based cryptosystem has many advantages,such as low computational complexity in encryption process,combined compression-encryption and robustness in decryption process.Therefore,CS has promising prospects in the application of information protection.However,the security level of standard CS framework is not satisfactory.The related research indicates that the standard CS framework cannot achieve the perfect security defined by Shannon.Therefore,it is necessary to perform further study on both the security of CS framework and its secure applications.This thesis mainly studies the applications of CS to image protection.One of the typical applications of image protection based on CS is to design image encryption algorithm using CS.Since CS based image encryption algorithm can satisfy the requirement of simultaneous encryption and compression,the designing of this kind of algorithm has attracted much attention.However,as image is a kind of multimedia data with bulk information capacity,the efficiency of CS sampling and reconstruction should be taken into consideration when encrypting image using CS.Besides,the standard CS framework cannot fully protect the image.Therefore,one of the specific works in this thesis is to design a secure and efficient CS based image encryption algorithm by combining some other cryptographic modes with the CS framework.Another attractive security application based on CS is to apply CS framework to cloud computing environment for image privacy protection and management.In this kind of application model,on the one hand,the security of CS framework could fulfill the privacy preserving requirement of secure computing in cloud;on the other hand,the cloud with abundant computing resource could handle the problem of high computational complexity in CS reconstruction.In other words,CS framework and cloud computing service are complementary to each other.Moreover,some fundamental signal processing can be performed in the CS domain,provided random matrix is used as the CS measurement matrix.This fact makes it possible to manage privacy preserving data in cloud environment.Because of the harmonious relationship between CS framework and cloud computing,it is not hard to see that the CS based information protection service in cloud environment has its promising prospect.Therefore,focusing on the topic of CS and its application in image protection,this thesis has done the following work:(1)In the respect of theoretical analysis on the security of CS,we summarize the security strength of the CS and analyze its security threat.The results provide some theoretical guidelines for designing image protection application based on CS.(2)For the application in image information protection using CS,we propose an image encryption algorithm by introducing diffusion mechanism in the parallel CS framework.This algorithm addresses the efficiency problem when encrypting image using CS.Besides,this algorithm also overcomes the defect of energy information leakage in CS based cryptosystem.(3)On designing the image encryption algorithm based on CS,we propose an image encryption algorithm combining the optical encryption technique and CS for the purpose of preserving the robustness of CS.This algorithm preserves the decryption robustness of CS,while ensuring the security and efficiency of the whole cryptosystem.(4)On designing the application combined with CS and cloud computing,we propose a privacy preserving CS image storage model in cloud environment.The proposed model can efficiently address the problem of high computational complexity in CS reconstruction by exploiting the abundant computing resource of cloud server,while simultaneously fulfill the requirement of privacy preserving in secure cloud computing.(5)Based on the proposed image storage model in cloud,we further design a privacy preserving image copyright authentication model by using the technique of watermark detection in CS domain.The proposed model effectively addresses the difficult problem of managing the privacy preserving data in cloud environment.Hence,it extends the application of CS based image protection service in cloud to a broader scenario.