Video Mining for Observing Human Activities

With the advance in video technology, video cameras have become an integral part of daily life. They are installed in parking lots, traffic intersections, airports, banks, etc. Usually a human operator watches them to catch events of interest in the scene, but this is a tedious and time consuming process requiring constant attention, and leads to inadequate surveillance capability. Therefore, there is an urgent need for automated systems for analysis of surveillance video streams.

This thesis presents a novel operational computer vision framework for visual knowledge extraction from human motion. The system captures a video of a scene and classifies those moving objects which are characteristically human. Then perform analyzing and mining operations based on full frame based analysis and inter frame based analysis to interpret the current activity. Moreover, based on selective criteria from full frame board and inter frame board the system evaluate the current activity to assist the security officers to catch the events of interest moreover, creating multi storing scheme for reducing the storage capacity in 24 hours surveillances system.