Grant: Discovery Grants Program - Individual

Camera networks are an emerging enabling technology for a broad range of applications, including video surveillance, environment and traffic monitoring, elderly care, intelligent environments, and urban and participatory sensing. Most researchers who are motivated by the challenging open problems in camera network research, however, simply do not have the resources to work with a large-scale physical camera network, given the cost and other serious impediments of deploying and experimenting with a camera network of appropriate complexity in an extended public space. A compelling alternative is ``Virtual Vision'', a novel camera network research paradigm that prescribes using visually and behaviorally realistic virtual environments for developing and evaluating camera networks. Virtual vision enables researchers to deploy virtual camera networks in visually and behaviorally realistic 3D computer simulated environments and to experiment with these networks under realistic conditions. In prior work, I have implemented a prototype virtual vision simulator and exploited it to develop two novel camera networks that are capable of carrying out observation and sensing tasks with minimal human intervention. I now propose to start a new research program at the UOIT, whose objective will be to take this promising initial work to the next level. The proposed research program will have the following four objectives: 1) to implement richer 3D simulations for virtual vision by developing new, more powerful behavioral and cognitive virtual human models, 2) to develop scalable 3D simulations of this kind on high performance computing systems, 3) to develop next-generation ad hoc networks of smart cameras by tackling fundamental problems in multi-camera control and coordination, and 4) to validate the virtual vision paradigm by using a real, physical multi-camera system.