Anamoly detection for safety monitoring

In crowded scene abnormal event detection is a major issue. Many existing methods are there. Abnormal events are those which cannot be well represented. For example, if a flight is hijacked or it is damaged, it is due to some abnormal activities. Abnormal activities may occur due to human intervention or due to some weather conditions. So in this system we are using abnormal detector to detect the events. Abnormal patterns are extracted from incoming events. The major contribution to this paper are: 1) In this abnormal detector is used to identify abnormal events. In this complexity is high in video events due to the presence of noise. By using mixture of Gaussian interference can be avoided. 2) In this, we are using Gaussian Mixture Model to reduce interference. Even though the method has high complexity. 3) Unusually normal events occur in testing videos which differ from training once this is due to existence of abnormalities. They presented as an online updating strategy is proposed to cover these cases in normal patterns as a result, it mostly eliminates false detections. Effectiveness of the proposed algorithm Is verified by using state of the art.