Virtual Vision and Smart Camera Networks

This paper presents camera network research in the context of a unique synthesis of advanced computer graphics and vision simulation technologies. In particular, we propose the design of and experimentation with simulated camera networks within visually and behaviorally realistic virtual environments. Specifically, we demonstrate a smart camera network comprising static and active simulated video surveillance cameras that provides perceptive coverage of a large virtual public space, a train station populated by autonomously self-animating virtual pedestrians. The readily reconfigurable virtual cameras generate synthetic video feeds that emulate those generated by real surveillance cameras monitoring public spaces. With minimal reliance on a human operator, our new camera network control strategy enables the collection of cameras to collaborate in performing various visual surveillance tasks, such as closely monitoring a pedestrian as (s)he locomotes in and out of the fields of view of individual cameras. The camera network supports task-dependent node selection and aggregation through local decision-making and inter-node communication. We propose a solution to the node selection problem that transforms it into an equivalent constraint satisfaction problem. Our technique is scalable and robust to node failures, as it lacks any central controller. Given the challenges of carrying out such research on a large scale in the real world, our simulation approach, which runs on a highend commodity PC, has proven to be beneficial for development and experimentation.