POP: Privacy-Preserving Outsourced Photo Sharing and Searching for Mobile Devices

Abstract

Facing a large number of personal photos and limited resource of mobile devices, cloud plays an important role in photo storing, sharing and searching. Meanwhile, some recent reputation damage and stalk events caused by photo leakage increase people's concern about photo privacy. Though most would agree that photo search function and privacy are both valuable, few cloud system supports both of them simultaneously. The center of such an ideal system is privacy-preserving outsourced image similarity measurement, which is extremely challenging when the cloud is untrusted and a high extra overhead is disliked. In this work, we introduce a framework POP, which enables privacy-seeking mobile device users to outsource burdensome photo sharing and searching safely to untrusted servers. Unauthorized parties, including the server, learn nothing about photos or search queries. This is achieved by our carefully designed architecture and novel non-interactive privacy-preserving protocols for image similarity computation. Our framework is compatible with the state-of-the-art image search techniques, and it requires few changes to existing cloud systems. For efficiency and good user experience, our framework allows users to define personalized private content by a simple check-box configuration and then enjoy the sharing and searching services as usual. All privacy protection modules are transparent to users. The evaluation of our prototype implementation with 31,772 real-life images shows little extra communication and computation overhead caused by our system.