A study on human abnormal activity detecting in intelligent video surveillance

Human abnormal activities detection in video surveillance is the most attractive topic of computer vision. In automated video surveillance systems, the human activities are monitored and then the system will generate an alert when any anomalous activity happens. System is modeled to acquire patterns of normal/abnormal behaviors using the different methods of human activity detection/recognition. An event/activity that differs from a normal pattern is identified as anomaly. Research and approaches in video surveillance covers extensive range of applications. An overview of various forms of anomalies, different kind of features and the existing machine learning approaches which are reported in the literature are reviewed. Few future directions are also highlighted for further research.