After defining a solid base for video pre-processing and human tracking,
the alarm detection can be faced with a high guarantee of
success in our expert video-surveillance system for shopping malls.
In this context, the objective is to give off alarms which allow to
detect certain risk situations or suspicious behaviors inside the
stores. As shown in Fig. 6, the high-level view of the proposed system consists of a multi-camera approach where three differentiated shop zones are under monitoring. Each zone has a particular
situation to evaluate: entry or exit of people in the entrance zone,
loitering events in the shop interior and unattended cash desk situations in the payment area. The video information from the IP
cameras in the shopping malls is sent employing a Web-based
communication, and the data loss and pixelation effects derived
from low bandwidth circumstances must be taken into account.
The video data is processed depending on the camera situation
and the alarms which must give off. Finally, the detected alerts
are presented to the human operators in the control center, who
can manage a higher number of cameras than using traditional
video-surveillance systems.