A Survey on Data Mining Techniques for Surveillance of Real Time Video Streams

Video Surveillance is the system that includes monitoring of the behavioural activities, or other changing information in the videos, usually of people for the purpose of protecting, influencing, directing, and managing. The development video of surveillance systems have recently captured the interest of both research and industrial worlds due to abnormal instances in industries, society and the increasing safety and security concerns. But it is not possible to monitor the videos twenty four hours for human beings. So there is need of such a system which will be useful for real time automatic object identification, object movement pattern recognition, modelling and detection of normal and abnormal (interesting) events, and recognition of event pattern. Data mining is the application of statistical techniques and programmatic algorithms to find out previously unknown relationships within the available data. On the basis of clustering and modelling normal events a new proposes algorithm will find out whether a segment has normal or unusual events. The existing techniques also compute the degree of abnormality of segment in addition to deciding whether the event is normal or unusual. The degree of abnormality means that to what extent a segment is differ from existing segments in relation with normal events.