TelosCAM: Identifying Burglar Through Networked Sensor-Camera Mates with Privacy Protection

Shaojie Tang,Xiang-Yang Li,Haitao Zhang,Jiankang Han,Guojun Dai,Cheng Wang,Xingfa Shen

DOI: https://doi.org/10.1109/rtss.2011.37

2011-01-01

Abstract:We present TelosCAM, a networking system that integrates wireless module nodes (such as TelosB nodes) with legacy surveillance cameras to provide storage-efficient and privacy-aware services of accurate, real time tracking and identifying of the burglar who stole the property. In our system, a property owner will have a wireless module node (called secondary module) attached to the property that s/he wants to protect. The secondary wireless module node will not store any personal information about the owner, nor any specific information about the property to be protected. Each user of the system will also have a unique wireless module node (called primary module) that contains some security information about the user, thus should be privately held by the user and be kept to the user always. Once a tracking process is triggered in privacy preserving manner, the secondary module will start sending out the alarm signal periodically. The alarm signal will be captured by some surveillance wireless module, integrated with existing surveillance cameras. Using the trajectory information provided by the secondary wireless module node, and the videos captured by the surveillance cameras, our system will then automatically pinpoint a burglar (e.g., a person or a car) that is more likely to carry the stolen property. Our extensive evaluation of the system shows that we can find the burglars with surprisingly high accuracy under various experiment settings, with significantly reduced storage-requirement of the legacy video surveillance system. It also can help the police to catch the burglars more efficiently by providing critical images or videos containing the burglars.