Research On Dynamic And Verifiable Encrypted Image Retrieval

With the rapid development of cloud computing and the explosive growth of images,people are outsourcing image data to cloud servers to save local storage resources and computational overhead.In order to ensure the security of the image and prevent privacy leakage,the image owner chooses to encrypt the data before uploading it to the cloud server.Image retrieval schemes based on searchable encryption technology can achieve searchability while ensuring the security of image data,but most encrypted image retrieval schemes ignore the threat of malicious cloud servers,which may return incorrect or incomplete search results.Since it is difficult to construct a common authentication structure to verify the similarity calculation process of image data,the verification of the encryption image retrieval process and the search results face great challenges.This paper mainly studies the malicious cloud server problem faced in the process of encrypted image retrieval,and the problem of image owner’s management of dataset.The main research works of this paper are as follows:(1)A dynamic and verifiable encrypted image retrieval scheme based on a ball tree is proposed to address the problems that the cloud server may return incorrect or incomplete results and the scheme does not support dynamic update of the dataset by the image owner and low retrieval efficiency.Firstly,the scheme uses a pre-trained CNN model to extract image feature vectors to improve retrieval accuracy.Secondly,a cryptographic index based on ball tree is designed to improve the retrieval efficiency.Then,the chameleon hash function is effectively combined with the Merkel hash tree to construct the chameleon hash authentication tree,which allows image owner to update the image dataset at a lower computational cost.Finally,the ball tree constructed in the scheme with the chameleon hash authentication tree structure allows query users to verify the retrieval process and retrieval results.The security analysis and experimental results demonstrate the correctness,security and high retrieval efficiency of the scheme.(2)A dynamic and verifiable encrypted image retrieval scheme supporting multi-user queries is proposed for existing schemes mainly applicable to single-user scenarios,while in reality people often use cloud storage to share image data with other users.Firstly,the scheme uses the secure kNN method based on LWE to encrypt the index and query vectors,which ensures the privacy of the index and query and the unlinkability of the query.Then,the proxy re-encryption technology is introduced,and the cloud server uses the re-encryption key to re-encrypt the retrieved ciphertext images,so that each query user can decrypt the ciphertext retrieval results without sharing the key with the image owner,and can use its own private key to decrypt the ciphertext retrieval results,so as to meet the needs of multiple query users to retrieve image data efficiently and safely.The security analysis and experimental results demonstrate the correctness,security and high retrieval efficiency of the scheme.