Based On The Cloud Image Encryption Security Technology Research

With the advent of the digital age,image data has been growing rapidly and image plays an important role in our daily life.In medical,financial and other fields,workers often need to store,forward and browse some images containing sensitive information.However,under the condition of limited local storage resources,the storage and security management of these images is a great challenge for users.With the emergence of cloud computing technology,third-party cloud storage is widely welcomed.However,as a third party cloud server is not completely trusted,the image information uploaded by users may be leaked to the third party server,resulting in the disclosure of user information privacy.In the cloud storage mode,the user's image data are all outsourced to the cloud server for unified management and management.This makes data security and privacy become the primary issue in cloud storage security technology.In order to protect the privacy of the image information and prevent the image from being illegally accessed,the image should be encrypted before the user uploads the image data to the server.However,after the image is encrypted,the correlation between the image data and other characteristics will disappear,which hinders their use of the image and does not meet the actual needs.And with the increase of the number of image storage in the server,the difficulty for users to get useful information from the image library also increases.Therefore,how to retrieve images under the premise of protecting image privacy has become a hot issue in this field.Aiming at the above problems,this paper studies the image retrieval scheme based on ciphertext domain in cloud environment.Its main work is as follows:(1)An efficient symmetric homomorphic encryption scheme is proposed,and the histogram,signal-to-noise ratio,correlation analysis and other experimental simulation results show that the encryption scheme has good security.(2)By combining the SIFT feature extraction,K-means clustering algorithm and the symmetric homomorphic encryption scheme designed above,an image retrieval scheme on the ciphertext domain is designed,and two servers are used to carry out image feature extraction and image retrieval simulation.The experimental simulation under Corel-1000 data set and Oxford Buildings data set shows that this scheme can realize image retrieval in encrypted images.Compared with the original scheme,the feature reduction rate of the proposed scheme in ciphertext domain is more than 90%,which is 2% higher than that of the original scheme.The accuracy of feature matching is stable at about 90%,and the matching time is increased by 60% compared with the original scheme.This scheme can save up to 30% in indexing time,and the retrieval time is equivalent to that in plaintext domain.The retrieval accuracy is over 90%,which is 1.5% higher than the original scheme.