A Privacy Preserving Video Surveillance System for Trauma Rooms

Abstract

Video surveillance in trauma room can improve the security, patient safety, healthcare standards, and hygiene monitoring, but video surveillance system can cause privacy breach by showing and storing sensitive information of patients and staff. Head blurring can protect privacy, and we can still able to know most actions of people with blurred head. But when postures of people in the videos are complex or when there are many people, using real-time detection program to detect and blur heads is challenging, with a high missed detection rate. OpenPose is a state-of-the-art human body skeletons estimation framework. And we build a head detection and blurring technique based on OpenPose to solve this problem. We first use OpenPose to extract 18 human key skeleton points, and find 5 key skeleton points which are used to describe head. Finally, we apply the image obfuscation techniques to the detected heads. The experiment results show that our proposed techniques can ensure privacy protection and reduces the missed detection rate of the output results. Our experiment results prove that our proposed technique has a high value in privacy protection applications.