| The purpose of this research is to facilitate the design of cost effective camera networks. It begins by identifying the design parameters and functional capabilities common to any type of multiple camera system. Generally, these attributes refer to the physical layout of the network, its topological organization, and the ability to which it can characterize the local environment. The performance of these abilities as a function of network size---i.e., the number of cameras---is investigated through computer simulation. In addition, possible correlations between these abilities are examined, as a single design parameter may affect the performance of several attributes simultaneously. The typical requirements of an automated refereeing scenario are identified and potential designs are investigated. An inward-looking network of twelve cameras is constructed and used for experimentally verifying the results of earlier simulated predictions. Ideally, by characterizing a camera networks abilities as functions of network size, an efficient design can be conceived for difficult scenarios, where a minimum level of redundancy is required to ensure balanced performance across all capabilities. (Abstract shortened by UMI.)... |
