Video based traffic data ascertainment in the field

Dresden / Siegen, 11/04/2013 – As part of research projects a video detector for the detection of traffic data was developed at the Fraunhofer IVI that is gradually transferred from the pilot phase to the production phase. In cooperation with the firm avinotec initially 46 locations with 92 camera systems been installed at North Rhine-Westphalia highways with integrated detection algorithm since the beginning of this year. The data serve the local traffic operator to assess the traffic situation and for the derivation of traffic forecast.

Single camera images at discrete points in time are serving the detection of vehicles and the determination of traffic characteristics (traffic volume, speed, occupancy rate), supplemented by predefined detection-zones – mainly according the individual lanes.

For each image the section of the detection zone is dissected and a mean color value is determined per cell. An adaptive background estimation separates active cells, as parts of potential objects and inactive cells, as for further processing negligible backgrounds. Cohesive groups of active cells are forming segments, which normally correspond vehicles. If two identical or sufficiently similar segments in two successive camera images are identified, they are associated with each other and linked to a movement. The counting of vehicles occurs when passing over a detection range of a previously defined counting line.