Knowledge Extraction from Surveillance Sensors
Rama Chellappa, Ashok Veeraraghavan, Aswin C. Sankaranarayanan · Wiley Handbook of Science and Technology for Homeland Security · 2008
Abstract During the last decade, significant progress has been made in both the design of novel sensors and in the development of algorithms for extracting information from these sensors. Several applications have emerged in areas such as automated wide area surveillance, monitoring, multimedia forensics, and video indexing and retrieval. Such systems use a plethora of sensors including motion sensors, audio sensors, visual and infrared (IR) sensors, along with accompanying algorithms to extract knowledge from the data captured by these sensors. Specifically, in the context of surveillance, the sensors mentioned above can effectively accomplish several tasks such as detecting and tracking humans and vehicles of interest, identifying human subjects, classification of vehicle types, human action analysis, and alerting security personnel when suspicious activities such as incursions into restricted areas are detected. In this article, we briefly describe the various sensors that can be used in such a system and discuss their advantages and disadvantages. We then discuss several classes of algorithms that are used in order to detect, track, and recognize people and vehicles in wide areas using a combination of both visual, audio, and motion sensors. Finally, we provide an overview of the current challenges and future trends.