Main Article Content
Abstract
Human fight detection is a considerably important topic for video surveillance, and for strengthening automated security systems. Various methods have been used by researchers to detect the fight or violence in surveillance systems. Previously proposed methods do not support the efficiency-complexity trade-off. In In this paper, Kalman filter and Blob analysis are used to detect the fight of multiple persons in video surveillance. First of all, objects are created for reading the videos. Then, foreground detection and blob analysis are used to find out the important characteristics like area, bounding box etc. After that structure is used to keep the state of tracking the object. Kalman filter is used to detect and track the motion of objects. It takes the Euclidian Distance between the predicted center and center of finding. This method works well when the camera is stationary. This method has been applied to UCF crime dataset and significant accuracy has been achieved.