Behavior Recognition System for Identifying and Monitoring Human Activities

Differentiating between normal human activity and suspicious behavior is a difficult task, wether performed by a sensing device or a human observer. Such a human observer would find such a task tedious and costly to perform. Fortunately, a sensing system is not bothered by tedious tasks. Such a system could be implemented to prume out obviously normal behavior, and tag human activities which could be suspicious and would therefor need more attention by a human operator. However, such behavior recognition systems have not been developed due to the difficulty of identifying and classifying such motions. We propose the development of a system which not only identifies humans in the environment and their location, but can also classify and identify their activity, providing a threat assessment. The heuristics needed involve recognition of information bearing features in the environment and the determination of how those features relate to each other over time. This is gesture recognition. This proposal addresses the technology development necessary to create a gesture recognition sensor system, enhancing it to create a behavior recognition, which would perform the dual task of classifying objects in terms of threats as well as determining the behavior (state) of objects in motion.