Behavior tracking model in dynamic situation using the risk ratio EM

摘要

Closed Circuit Television (CCTV) system has been popular in daily life such as traffic, airport, street and public place. The common goal of CCTV system is the prevention of crime and disorder by observing objects. In the future, smart CCTV camera combined with mobile phone will be used to protect human from crime and dangerous situations. Intelligent CCTV system in public place will monitor human behavior in real-time and transfer image data to control tower for the security purpose. In this paper, we propose an abnormal behavioral tracking model for prediction of abnormal situation by using Expectation Maximization (EM) algorithm combined with Viterbi algorithm. The tracking model will detect objects from CCTV image in dynamic environment for the prediction of dangerous situation. This tracking system has five main steps. (1) The detection of object and their environment, (2) Feature extraction from objects and situations such as human body posture, weather, and time (3) Location information such as object trajectory and area safety level (4) knowledge update and decision making (5) prediction of abnormal situation and maximized risk rates.