Suspicious Activity Recognition in Video Surveillance System

In today's insecure world the video surveillance plays an important role for the security of the indoor as well as outdoor places. The components of video surveillance system such as behavior recognition, understanding and classifying the activity as normal or suspicious can be used for real time applications. In this paper the hierarchical approach is used to detect the different suspicious activities such as loitering, fainting, unauthorized entry etc. This approach is based on the motion features between the different objects. First of all the different suspicious activities are defined using semantic approach. Then the object detection is done using background subtraction. The detected objects are then classified as living (human) or non living (bag). These objects are required to be tracked which is done using correlation technique. Finally using the motion features & temporal information the events are classified as normal or suspicious. As the semantic based approach is used computational complexity is less and the efficiency of the approach is more.