Abstract

Indoor physical security, as a perpetual and multi-layered phenomenon, is a time-intensive and labor-consuming task. Various technologies have been leveraged to develop automatic access control, intrusion detection, or video monitoring systems. Video surveillance has been significantly enhanced by the advent of Pan-Tilt-Zoom (PTZ) cameras and advanced video processing, which together enable effective monitoring and recording. The development of ubiquitous object identification and tracking technologies provides the opportunity to accomplish automatic access control and tracking. Intrusion detection has also become possible through deploying networks of motion sensors for alerting about abnormal behaviors. However, each of the above-mentioned technologies has its own limitations. This thesis presents a fully automated indoor security solution that leverages an Ultra-wideband (UWB) Real-Time Locating System (RTLS), PTZ surveillance cameras and a Building Information Model (BIM) as three sources of environmental data. Providing authorized persons with UWB tags, unauthorized intruders are distinguished as the mismatch observed between the detected tag owners and the persons detected in the video, and intrusion alert is generated. PTZ cameras allow for wide-area monitoring and motion-based recording. Furthermore, the BIM is used for space modeling and mapping the locations of intruders in the building. Fusing UWB tracking, video and spatial data can automate the entire security procedure from access control to intrusion alerting and behavior monitoring. Other benefits of the proposed method include more complex query processing and interoperability with other BIM-based solutions. A prototype system is implemented that demonstrates the feasibility of the proposed method.