Adaptive CCTV display aims to allow operators to detect incidents more effectively using existing CCTV cameras. Alerts from CCTV automatic incident detection (AID) systems are combined with live data from inductive loops and historic data to identify sections of the road network where incidents may have occurred. Images of these discrete sections of road are then displayed to operators. Using AID on existing pan-tilt-zoom (PTZ) cameras typically gives an unacceptably high false-alarm rate. This is mitigated by probabilistic data-fusion and an effective but unobtrusive display. A prototype system has allowed operators to detect incidents around three minutes faster than without the system. Work to enhance the system will increase its long-term maintainability, improve its user interface, and enhance the data fusion algorithms at its core. These data fusion enhancements will see the system make use of richer information from underlying AID systems and present to users with more information on the likely cause of an alert.
