Scheduling an active camera to observe people
- 15 October 2004
- proceedings article
- Published by Association for Computing Machinery (ACM)
Abstract
Remote identification of people is an important capability for security systems. Automatically controlling a pan-tilt-zoom camera is an effective way to collect high resolution video or images of people in an unconstrained environment. Often there will be more people in an area than cameras available. The cameras must then divide their time among the people in order to view everyone. In this paper, we discuss the challenges involved in scheduling an active camera to observe multiple people. We present some candidate scheduling policies to address these challenges and evaluate their performance. The evaluation was conducted with a simulation based on data collected with our cooperative active camera system.Keywords
This publication has 8 references indexed in Scilit:
- Real-time object classification and novelty detection for collaborative video surveillancePublished by Institute of Electrical and Electronics Engineers (IEEE) ,2003
- A master-slave system to acquire biometric imagery of humans at distancePublished by Association for Computing Machinery (ACM) ,2003
- Intelligent environments and active camera networksPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2002
- Adaptive Resolution System for Distributed SurveillanceReal-Time Imaging, 2002
- Scheduling Multiclass Packet Streams to Minimize Weighted LossQueueing Systems, 2002
- Approximating the Throughput of Multiple Machines in Real-Time SchedulingSIAM Journal on Computing, 2001
- Learning patterns of activity using real-time trackingPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2000
- On-line schedulingPublished by Springer Nature ,1998