| Abstract |
With recent advances in computer vision techniques, it is now becoming possible to look for high-level semantic information in video streams. Computer vision systems are becoming more and more popular in monitoring and surveillance applications. Such systems can be used for the surveillance of public places, banks, parking areas, airports, highways for traffic monitoring, etc. The development of vision-based monitoring and surveillance systems requires the capability of detecting interesting events in a video stream. Towards this goal, most available systems rely on the detection, recognition and tracking of real world objects. The proposed approach decouples the detection and interpretation of events from the explicit, computer-based detection and recognition of objects, actions, and their evolving relationships. This is very important because it facilitates the deployment of a system in a variety of application domains. Additionally, the proposed approach requires very simple low-level vision processes, which leads to robust and efficient performance. The proposed approach is based on the use of logical sensors defined over specific regions of the image. Logical sensors decide whether a certain property holds in a specific image region at a specific moment in time. Logical sensors, in fact, enable the detection of primitive events in a video stream. On top of these elementary sensors, temporal and logical aggregation mechanisms are used to define hierarchies of progressively more complex sensors, able to detect events with more complex semantics. Finally, scenario verification mechanisms are employed to achieve process monitoring, by checking whether events occur according to a predetermined order. The proposed approach was implemented and has been tested and validated in an application involving the monitoring of automated processes. The obtained results demonstrate that the proposed approach provides a promising framework for vision based event detection.
|
Ανάπτυξη οπτικού συστήματος ανίχνευσης συμβάντων και επόπτευσης διαδικασιών για εφαρμογές βιομηχανικού αυτοματισμού
