Efficient privacy-preserving motion detection for HEVC compressed video in cloud video surveillance
Xiaojing Ma, Bin B. Zhu, Tao Zhang, Sixing Cao, Hai Jin, Deqing Zou · 2018
Privacy-preserving information utilization has attracted a lot of attention in the big data era where the data owners and data processers may be separate entities. Cloud video surveillance is a typical example that requires privacy-preserving data processing. Video surveillance can leverage the cloud for its storage, processing power, and accessibility from anywhere with any devices. In such a system, surveillance cameras simply stream video data to the cloud, which then stores the video data, performs motion detection and other analysis, and alerts users if needed. Cloud video surveillance presents a unique challenge: privacy protection of surveillance video since it will be transmitted over public networks, stored and possibly processed at a third-party cloud that may not be trusted. This paper addresses this challenge by proposing a method that allows a third party to perform motion detection directly on the encrypted and compressed surveillance video without decryption. Specifically, we propose the first compressed-domain privacy-preserving motion detection method that preserves the compression efficiency of HEVC compressed video. The proposed method can detect the coarse-grained shapes of moving objects and then estimate the motion trajectory. In addition to the preserved high compression efficiency of standard video coding, the proposed method also has the following desirable properties: 1) it only needs one cloud to participate; 2) authorized users can fully recover the original compressed video recorded by the camera; 3) low computational complexity.