LHMM-based gathering detection in video surveillance

Automatic detection of unusual event in video sequence has an interesting application in security surveillance. This paper proposed a method to detect a gathering event without tracking or analyzing individual activities. We divide the video scene into blocks and rely on background subtracted blobs and optical flow instead of tracking statistics as the features to extract information from the video data. The features are encoded with Layered Hidden Markov Models to allow for the detection of gathering event. The experimental results show the effectiveness of the proposed approach.