A multi-agent architecture to support active fusion in a visual sensor network

One of the main characteristics of a visual sensor network environment is the high amount of data generated. In addition, the application of some process, as for example tracking objects, generate a highly noisy output which may potentially produce an inconsistent system output. By inconsistent output we mean highly differences between tracking information provided by the visual sensors. A visual sensor network, with overlapped field of views, could exploit the redundancy between the field of view of each visual sensor to avoid inconsistencies and obtain more accurate results. In this paper, we present a visual sensor network system with overlapped field of views, modeled as a network of software agents. The communication of each software agent allows the use of feedback information in the visual sensors, called active fusion. Results of the software architecture to support active fusion scheme in an indoor scenario evaluation are presented.