Computer Vision vs. Traditional Video Motion Detection (VMD)
Traditional video motion detection (VMD) must not be confused with true, intelligent computer vision video analytics.
Traditional video motion detection carries an inherent flaw:
it assumes any pixel change is significant.
Imagine a camera on an outdoor train platform. Pixel changes will be caused by movement of people on the platform and trains on the tracks, as well as, wind, rain, shadows, and light reflections. VMD will alert on all changes, creating an unmanageable number of false alarms, and presenting a high likelihood that the event of interest – a person crossing the train tracks – would be missed by monitoring personnel.
Computer vision technology filters irrelevant information and relays only significant data for analysis.
Using computer vision in the train platform example would mean that irrelevant pixel changes caused by environmental factors and normal behavior would be ignored. Monitoring personnel would be alerted only when the event of interest occurred – a person crossing the train tracks – and could respond appropriately. Essentially, computer vision based video analytics are highly sophisticated and complex mathematical algorithms. The analytic algorithms available with Aimetis Symphony™, for example, are capable of distinguishing relevant activity in a scene when it occurs and then "remembering" the details of that activity so that it can be easily found during an investigation or so that it can provide intelligent information in the form of reports or graphs that can be interpreted by users.