Secure Image Retrieval over Untrusted Cloud Servers

140 Abstract-Security issue represents the main barrier facing the wide adoption of cloud computing. Encryption is the best method to mitigate users' concerns. However, this method makes searching the encrypted data a challenging task. Accordingly, several approaches have been proposed to enable searching the encrypted, remotely stored data without decryption. Till now, almost all these approaches are limited to handle text search but not multimedia search. In this paper, we propose an efficient scheme that provides content based search over encrypted image database. To do so, we utilize a locality sensitive hashing LSH method to build our searchable index. LSH index greatly enhances the system efficiency by returning the matching images in a ranked order with a minimum number of distance evaluations. For security purposes, we turn this index into a secure index to prevent the cloud server from learning any useful information from the contents of that index. Searchable index along with image collection are outsourced to the cloud server in their encrypted format. We provide several empirical experiments to illustrate the efficiency of our proposed scheme.