Research On Encrypted Image Search In Mobile Cloud Computing

With the rapid development of mobile Internet,the multimedia data(such as image)is exponentially increasing.It has brought severe challenges to data storage and processing for resource-constrained users.Cloud storage is an extension and development of the concept of cloud computing,which provides a viable solution for huge amounts of data storage.In order to reduce their consumption of storage resources,resource-constrained users upload local data to the cloud for storage.However,since the cloud server is not fully trusted,and once the data is uploaded it will cause the separation of ownership and management of data.Therefore,the data is usually encrypted by the user in the local operation in order to ensure the confidentiality of data,then stored in the cloud server in the form of ciphertext.So,this presents some new challenges to the efficient retrieval of data.Currently,how to achieve efficient retrieval of encrypted image is one of the hot topics in the academic and industrial community.Although some existing encrypted image retrieval schemes realize the image retrieval under the ciphertext,there are many problems such as information leakage,computational overhead and communication overhead.This makes the application range and the practicability of the scheme is constrained,especially for the resource-constrained mobile devices.Therefore,it is very significant to study the efficient encrypted image retrieval technology in mobile cloud computing.This thesis focuses on the problem of image retrieval under the ciphertext in mobile cloud computing.The main contributions of this thesis are as follows:(1)We propose a new encrypted image search scheme based on comparable encryption and CBIR in mobile cloud computing.This scheme can not only protect the privacy of feature vectors of the image,but also realize the comparison of feature vectors under the ciphertext.In addition,the proposed scheme only requires one round of communication between the user and the cloud server,which reduces the communication overhead.Then we adopt the two layer retrieval model for image retrieval in the scheme.Firstly,the range retrieval is performed based on the tree index structure,and then the feature vectors in the range are retrieved accurately.This makes the scheme have better retrieval efficiency and accuracy,and avoids the problem that the excessive computational overhead caused by a single image similarity measure.The main result has been published in the international journal Soft Computing(SCI source journal,impact factor: 2.472).(2)We not only analyze the security and efficiency of the scheme in theory,but also realize the image retrieval under the ciphertext scheme based on the “INRIA Holidays” image data set.The results show that the retrieval cost of the proposed scheme is approximately7.9381 n milliseconds for the randomly selected images in the experiment,where n is the number of images.Therefore,the proposed scheme has better retrieval efficiency and accuracy.In addition,compared with encrypted image retrieval scheme based on homomorphic encryption algorithm,the computational overhead of the proposed scheme can be effectively reduced,so it can be better applied to resource-constrained environment.