Research On Block-level Deduplication For Encrypted Images In Cloud Computing

With the rapid development of Internet technologies and cloud storage services,the multimedia data such as images have been increasing explosively,and the storage costs and expenses have proliferated.At the same time,the high similarity and repetition of images result in the high redundancy of the stored data and serious waste of storage space.To solve the problem of image storage performance imbalance,image deduplication has become a research hotspot of image processing direction.However,the openness of the cloud may cause leaking of sensitive information,and cloud storage may lead to the separation of data ownership and management.In order to protect the security of images,encryption becomes an inevitable demand and trend.However,encrypted images are difficult to calculate and recognize by using traditional image processing algorithms.Encrypted image deduplication cannot directly use traditional deduplication methods.Therefore,the encrypted image deduplication scheme in the cloud environment is currently being extensively studied.The traditional accurate image deduplication scheme did not consider the difference of deduplication effect on different types of data.The deduplication method by checking the hash value of files makes a large number of similar images cannot be effectively processed.At the same time,the fuzzy image deduplication mechanism relies on the feature vector,which leads to the possibility of mistaken or insufficient deletion of images.To overcome the contradiction between image deduplication and encrypted image processing,the existing image deduplication schemes generally adopt techniques such as convergent encryption or message lock encryption.However,there are still security threats such as offline brute force attack or online brute force attack.Therefore,to solve the above problems,it is urgent to study how to implement the encrypted image deduplication scheme with deduplication precision configurability and higher security in the cloud computing environment.In summary,this thesis is based on the pixel blocking method and password authenticated key exchange protocol to study the security encrypted image deduplication technology in cloud environment.The main study contents are as follows:1.First,in order to solve the problem of the redundancy between different images,this thesis designs a block-based deduplication scheme for encrypted images which bases on a new pixel blocking method,and utilizes the featured hash value of image block to deal with the duplicate check.This scheme can identify the image blocks with the same visual perception,achieving the deduplication of the redundant blocks in different images and greatly improving the image deduplication rate.At the same time,by introducing the group key and encryption algorithm,the scheme can protect the content of the image blocks and the feature vectors.This work has applied for a national invention patent.2.Then,considering the existing schemes still face security threats such as offline brute force attack and online brute force attack,this thesis proposes a block-level image deduplication scheme for multiple users based on PAKE protocol.The scheme utilizes the new image blocking method to improve the image deduplication rate.In this scheme,users can choose either accurate deduplication or near identical deduplication according to their own settings,which ensures the interests of users who do not like the loss of any pixel.Meanwhile,this scheme has no restrictions on the type of image,and can be applied to most practical image libraries,which is more versatile than existing schemes.In addition,this scheme applies the PAKE protocol to the image deduplication scheme for the first time,which can resist threats such as online brute force attack and offline brute force attack.Compared with the existing encrypted image deduplication scheme,security has improved.This thesis conducted simulation experiments on both of the above scheme based on two image libraries,the performance analyses show the feasibility and effectiveness of the scheme.Related research can be applied on medical imaging and other fields,and beneficial to improve security and availability.