Research On Cryptography Of Privacy-preserving Image Retrieval In Cloud Environment

With the increasing use of multimedia services and applications by individuals and businesses,the storage and computing requirements of image data keep on growing in recent years,and the storage and management of large-scale image data have become an urgent problem to be solved.Cloud computing,an Internet-based computing method that could share hardware resources,software resources and information,can widely be supplied to the demand of various computer terminals.Due to the great flexibility and cost-saving features of cloud technology,users are increasingly willing to store image data from local devices to cloud servers.Although cloud computing technology can bring people a lot of convenience,when images contain sensitive information,storing images directly to the cloud server risks the disclosure of privacy,and cloud service providers can directly access and use the stored images.To protect images from privacy breaches,images are typically encrypted before being stored on the cloud server.However,the encryption operation will destroy the image content,which makes it challenging to perform image retrieval on the encrypted image.Some existing schemes can implement encrypted image retrieval,but there are still defects of low search efficiency and image information leakage.Recently,some attacks have been able to obtain user's search and image information according to the information leaked in the encryption scheme.This thesis considers the two aspects of establishing secure image index and fused image feature encryption,and studies the specific process of encrypted image retrieval.The main research work and contributions of this thesis are as follows:1.This thesis finds that the forward privacy can only partially solve the problem of search pattern leakage,and there is still the problem of leaking the users' search intention.Therefore,in this thesis,a multi-user image retrieval scheme that suppresses the leakage of search pattern is designed and implemented,and the problems of forward privacy and search pattern leakage are solved.The secure index tree is built by using chain blocks,which combines with search path shuffling to eliminate the search pattern leakage.Theoretical analysis and experimental results evaluate the oblivious of the scheme to suppress search pattern leakage,and the optimal search and update performance.2.Considering that image data has many characteristics,this thesis proposes an image retrieval scheme based on fused image feature encryption.In this scheme,due to the color information and texture information of the image can be separated,different encryption techniques are used to protect these features,thereby achieving less image privacy leakage.Since the texture information in the image is more important than the color information,this scheme selects the probabilistic encryption to encrypt the texture information to protect the image content and exploits the deterministic encryption to encrypt the color information.Finally,the feasibility of the scheme is verified by the experiments,and the security of the scheme is analyzed.