Research On Ciphertext Image Retrieval Scheme Based On Object Detection In Cloud Environment

With the advent of the era of big data and cloud computing,more and more enterprises and individual users choose to store a large number of image files in the cloud server,and provide image retrieval and sharing functions,which can not only save a lot of local space,but also provide efficient data retrieval and transmission services for data sharing between users.However,the security problem in the cloud server is becoming increasingly prominent.At present,the more recognized approach is to encrypt the data before it is uploaded to the cloud server.But the encrypted data can't be retrieved in the traditional clear text,and although the encrypted data can't be peeped by the cloud server,it can't prevent the cloud server from operating the data without the user's permission,that is,the integrity of the data may be damaged by the cloud.Therefore,it is of great significance to research and design a secure and efficient image retrieval scheme in the cloud environment.To solve these problems,this paper proposes a Ciphertext image Retrieval scheme based on Object detection in Cloud environment(CROC),which is divided into two schemes: Basic plan(CROC-B)and Expansion plan(CROC-E).In CROC-B,the deep learning model of object detection is used to preprocess the image.The key words set can not only roughly classify the image and speed up the retrieval efficiency,but also the generated image feature vector can more accurately express the image features compared with the traditional image processing method,so that the matching accuracy is higher.Then,the key words set is encrypted by multiple linear mapping By encrypting and building a security index,neither the attacker nor the cloud server can crack the encrypted content;CROC-E introduces blockchain technology on the basis of CROC-B,and the distributed system provided by blockchain technology improves the retrieval efficiency of image data.Its super ledger records image data information,which cannot be changed,for the integrity verification of image data.Security analysis shows that CROC-B can not only effectively resist the guess attack,performance evaluation shows that CROC-B has better retrieval efficiency and the highest retrieval accuracy compared with the existing scheme;CROC-E has faster retrieval efficiency compared with CROC-B in performance,and can provide relevant data for integrity verification for image data in security.