The safe operation of tunnels requires a fast and reliable incident detection. Only precise information about the place, type and extent of the incident allows for a selective launch of measures, e.g. regulation of the ventilation system for smoke control in case of fire, or the halting of traffic outside and inside the tunnel following an accident. At present, different types of detectors are deployed in tunnels to monitor the traffic area. The collection of traffic data is generally done through local measurement devices such as induction loops or overhead infrared sensors. Fire incidents are detected through fire detecting cables, fog sensors, or the removal of fire extinguishers from emergency stations. When video equipment is installed, these detectors are linked with cameras, so that an image from the incident spot is switched on in the permanently occupied control rooms. However, the fire events over the past years have shown that, often, fires were only detected by the optic sensors (fog sensors). Using the appropriate evaluation logarithms meanwhile, available computing capacities allow for an analysis of video data in real time and, at the same time, to automatically detect the type and extent of the incidents as well as their spatial and temporal causes. Apart from traffic data such as traffic flow, traffic density and speed, the following incidents can be detected automatically on the video image: formation of fire and smoke, stopped vehicles, use of hard shoulders resp. emergency bays, ghost driver, persons resp. objects on the road, changes in illumination. The online analysis of the video data then makes the direct and selective control of operating functions possible immediately following the incident occurrence. A further advantage of digital video analysis lies in the storage of recorded data, which can serve in the subsequent reconstruction of incidents. Digital video analysis allows for the integration of several types of detection within one system and can therefore be applied flexibly and universally. Consequently, apart from the pure collection of data, the entry of dangerous goods transports well as fires can be detected. Given a pertinent configuration of the cameras, a complete observation of the traffic area is possible. Since the analysis of data is done in real time, measures can be initiated immediately following the occurrence of an incident. Additionally, an on-the-spot video image is available at all times. Furthermore, owing to the camera arrangement in the ceiling area, no drawn-out interventions are necessary in case of maintenance or repair jobs to the road surface. (A) For the covering abstract of the conference see ITRD no 207828. The reprints are also available at the web - http://www.vti.se/pdf/reports/K18APart1.pdf; http://www.vti.se/pdf/reports/K18APart2.pdf and http://www.vti.se/pdf/reports/K18APart3.pdf.
Abstract