Abstract Virtual Vision Simulator For Smart Camera Network Research

We demonstrate a virtual vision simulator developed for the purpose of smart camera net-work research. The simulator uses the latest graphics technologies to produce a realistic looking scene complete with global illumination, shadows and photo-realistic pedestri-ans. We discuss our layered camera architecture that allows the cameras to complete high level tasks such as:“track the pedestrian with appearance signature h”. We also discuss our scripted pedestrians that provide full control over their position at any given time. To demonstrate our simulator, we show how the simulator can be used to research multitasking cameras as well as the use of a central planner for camera assignment. ii Acknowledgements I would like to thank Dr. Faisal Qureshi for giving me the opportunity to work with him, and for supervising me as I worked on my thesis. ii