Current surveillance and control system still require a human super-vision and intervention. India, which has only
three Lakh surveillance cameras under human observation, lags in the Surveillance System. This paper will present the capture
or observation of suspicious activities and warn the officer that immediate and sufficient action can be taken. Scenes are
translated into frames to capture an unusual scene and prepare data as an input for Algorithms. We will then define the frames
and translate them into vectors. Filtering a noisy data for each feature vector and if feature values deviate a threshold then it
detects a shot edge. Clustering an edge shot and constructing a scene with a limited number of frames. Furthermore, a limited
number of frames will reduce the cost of computing. This can detect suspicious objects and automate cameras from human
observation to machine vision, both public and private.
