Video Motion Detection (VMD) requires more computational power to be accurate. Detecting motion from a camera and being able to capture the moment in real-time, requires that various aspects to be taken into consideration. Some of these aspects are contrast created by weather outside, flying machines that may fly over a restricted zone and strong winds that can affect stationary objects to activate the system. There is a strong need from the security surveillance systems to detect and suppress these false alarms and to increase the sensitivity to significant events of interest.
This project focuses on motion segmentation, detection of moving targets, and dense motion field estimation to implement the solution with the minimum number of errors when detecting motion. The computational algorithm has to improve the speed of motion detection. One way of getting rid of false motion detection on a dynamic background is to process every single cell in order to avoid detection that set off by contrast level, small objects, and shifting of the camera’s position.