Research On Multi-feature Image Retrieval Algorithm Based On Privacy Protection

In the Internet age,with the rise of various social networks,pictures and video data in the network have grown at an alarming rate every day,gradually forming a powerful image retrieval database.For these massive images with rich information,how to effectively retrieve the images that users need from a huge image database has become a research direction of interest to researchers in the field of information retrieval.There are three main types of image retrieval algorithms: text-based,content-based,and semantic-based image retrieval.This paper mainly studies content-based image retrieval schemes(CBIR).In the early stage of the research,the researchers studied the image retrieval based on single feature.However,the accuracy of this method is low and the retrieval effect is not good.In order to further improve the retrieval robustness,multi-feature fusion image retrieval scheme has become a research hotspot.In order to more conveniently process large-scale image data,users need to store images in a cloud server.When the image owner outsources the CBIR service to the cloud server,the image owner loses control of the data and the data is vulnerable to common types of attacks during transmission.At this time,the transmitted data is easily acquired by a third party and in the ciphertext domain image retrieval scheme,the cloud server is in a semi-trust state.For this purpose,the image retrieval scheme under the ciphertext domain is studied.The current image retrieval scheme under the ciphertext domain needs to be further improved in protecting the data owner rights,feature extraction module and image encryption module.Based on the above considerations,this paper first studies the image retrieval scheme under the ciphertext domain,and its goal is to make up for the problem of insufficient data owner's authority.Secondly,a multi-feature fusion image retrieval scheme is proposed,which aims to improve the accuracy of image retrieval.Finally,the high-dimensional hyperchaotic system and DNA coding technology are studied to improve the security of outsourced CBIR services.The research content of this paper is as follows:An image retrieval scheme based on privacy protection is proposed.In image encryption,a 4-D hyperchaotic system and DNA encoding rules are used to generate hyperchaotic sequences,the goal of which is to protect data privacy.In terms of index encryption,a multi-feature fusion image retrieval strategy is proposed.In order to further ensure the security of the index,this paper uses the secure kNN algorithm to complete the index encryption.In order to improve the security access control authority,this paper proposes a re-authentication scheme based on the AES algorithm of OFB mode.After simulation testing,the proposed algorithm has better resistance to common types of attacks in this paper.In addition,the accuracy of the proposed algorithm is better than that of the single feature retrieval algorithm.The experimental results show that the proposed ciphertext domain image retrieval scheme has improved security and retrieval accuracy.It can resist common attack types such as differential attacks,plaintext attacks,noise attacks,and blocking attacks.