Video Sensor for the Acquisition of Road Traffic Data

For the automated, adaptive control of road traffic with light signal systems and dynamic route guidance systems, up-to-date traffic data are required in suitable quantity and quality.

Conventional induction loops deliver this data, but they are often too expensive for area-covering use. On the other hand, inexpensive systems based on Floating Car Data qualify for real-time application only when a significant minimum coverage of measurement vehicles has been reached. Video sensors can close the gap between those two approaches.

In addition to this, traffic control centers require that up-to-date images of congestion hot spots are provided as a standard.

For several years, the Fraunhofer IVI has been working on the combination of live camera systems and video sensors for the acquisition of road traffic data.

The result of this research is a software which automatically detects common traffic parameters in image sequences. The resolution and processing of images is carried out in due consideration of data privacy requirements. In particular, it is not possible to identify people or license numbers on vehicles.

The Fraunhofer IVI has developed a video sensor for the acquisition of road traffic data that runs on standard camera modules and computer systems. The system can take up to two camera modules which, depending on the field of view, record up to six lanes simultaneously. The sensor currently delivers the following traffic parameters:

• Traffic intensity
• Speed
• Density
• Length of traffic tailbacks and
• Type of vehicle (car/truck).

In addition to these parameters, further important information for the local control of traffic hubs will be made available, such as the detection of rucks and time gaps.

The sensor has been designed for typical viewing angles, for instance from rooftops or lamp posts, and it can be installed without major measuring of the traffic scene. Depending on geometrical and weather-specific visibility conditions, detection rates of more than 95 percent can be achieved for the measurement of traffic intensity.

The video detectors located all throughout the road network are connected to a central server by WLAN or data cable. This server can then provide measurement data via standard interfaces for any database or central traffic control computer. The next development step will connect the sensor directly to local control computers of light signal systems.

The development of the video sensor was co-funded within the research projects »intermobil« and »ORINOKO«.