Distributed Sensor Networks For Visual Surveillance

Automated video analysis systems consist of large networks of distributed heterogeneous sensors. Such systems require extraction, integration, and representation of relevant data from sensors in real time. This book chapter identifies some of those major challenges and proposes solutions to them. In particular, efficient video processing for high-resolution sensors, data fusion across multiple modalities, robustness to changing environmental conditions and video processing errors, and intuitive user interfaces for visualization and analysis are discussed. Enabling technologies to overcome these challenges are also discussed. The case study of a wide area video analysis system deployed at ports in the states of Florida and California, USA is also presented. The components of the system are also detailed and justified using quantitative and qualitative results.