A Survey of Intelligent Visual Surveillance Systems

This paper presents a survey of Intelligent Visual Surveillance Systems (IVSS) with the aim to identify the needs and requirements of future large-scale (IVSS). Pros and cons of existing IVSS are discussed from different perspectives including: motion segmentation, region of interest (ROI) characterization, region of interest tracking, event recognition, and scenario profiling. Most IVSS approaches have limited fusion focus, superficial, and lack sufficient cohesiveness and completeness. By introducing sensor data and information fusion techniques, on the other hand, one can significantly improve the critical decision making at different levels of information abstraction and hence reducing decision uncertainty and improving ROI profiling in context-sensitive environments. In this paper, we have systematically addressed technological needs of different approaches supporting IVSS and identified a future direction to further enhance the current state-of-the-art of IVSS for homeland security and battlefield applications.