When you are using a closed circuit TV system, in case of particular events it may happen to have to sit and closely watch hours and hours of footage that always looks the same, to find those few seconds in which something important is happening, and try to extract clues or information out of those scarce, and sometimes blurred images, while the risk of losing important details because of an understandable lapse of attention is always present.
In order to avoid losing details which might be of vital importance for any kind of investigation, the Hebrew University of Jerusalem has developed a software to analyze video surveillance footage.
This software quickly scans and analyzes the footage shot by a surveillance camera, for example during a night when something happened to the area under control, squeezing everything into a few minutes of footage which will contain only images showing events of relevance; all this is achieved thanks to an algorithm which can tell a still image from one which contains moving parts or objects.
In practice, what the software does is getting rid of all those hours in which absolutely nothing happens inside the camera’s visual range, saving only the moments in which cameras detect a motion. In this way, those who later analyze the footage can, in case they notice something important, return to the original video and view it normally to place the images in a context.
This kind of software might find a field of application for video surveillance of areas where nothing is supposed to happen during the night, for example like the dark halls of a museum, while it may not be very useful if used, for example, to monitor areas with a lot of traffic, for example transportation companies with cargo trucks moving around.