Research Of Motion Detection Technology On Encrypted Surveillance Video

Cloud-based video surveillance has attracted more and more attention for its ubiquity,scalability and robustness.However,videos in the cloud usually includes privacy of individuals and organizations which is faced with the danger of stealing.Therefore these videos should be protected from privacy leakage.One promising solution is to encrypt the video before it's uploaded to cloud.Besides it is a trend to use the computing power of cloud to process and analyze surveillance videos,and motion detection and tracking are main tasks to perform in a video surveillance system.Altogether,cloud-based video surveillance requires protecting privacy yet allows cloud to perform motion detection and tracking.Prior arts allow performing surveillance on encrypted videos but the video encryption and detection are all performed in pixel domain which is hard to meet the requirement of instantaneity.Recently,several new arts allow performing encryption and motion detection in compressed domain,but the details of trajectories of moving objects are completely exposed.Research of motion detection technology on encrypted surveillance video proposed a cloud-based video surveillance system based on H.265/HEVC standard that supports adjustable granularity of motion detection and tracking for fine-grained control on conflicting requirements between privacy protection and accuracy of motion detection.The proposed encryption adds a permutation layer which includes permutation of CTU and permutation of partitions inside CTU to conventional format-compliant selective encryption.The proposed motion detection method estimates both local motions and global motions to deduce trajectories of moving objects.The experimental results prove that the system has a good accuracy in motion detection and tracking and achieves controlling adjustable granularity of motion detection,and has a very small impact on video compression.