In video surveillance, detection of relocating objects from a video is foremost for object detection, goal tracking, and habits figuring out. Detection of moving objects in video streams is the first critical step of expertise and historical past subtraction is an extraordinarily general method for foreground segmentation. In this thesis, we now have simulated special background subtraction approaches to overcome the obstacle of illumination variant, historical past muddle and shadows. Detecting and monitoring of human physique elements is most important in figuring out human movements. Intelligent and automatic security surveillance programs have emerge as an lively study area in latest time because of an increasing demand for such systems in public areas equivalent to airports, underground stations and mass activities. On this context, tracking of stationary foreground regions is likely one of the most principal requirements for surveillance systems centered on the tracking of deserted or stolen objects or parked autos.
Keywords : object detection, tracking, Kalman filter, occlusion, image processing .