Lund University is developing an automated video analysis system for making practical studies in traffic environments. The system can be used on three application levels: I. Collection of the ordinary traffic data (flow, speed distribution, delays, stops) II. Selection and description of particular events (e.g. encounters between a pedestrian and a car at a pedestrian crossing). III Rare events detection — “watchdog” (Conflict studies, incident warnings). The system is primarily designed for complex urban environments and can handle various road user categories, i.e. not only motor vehicles, but also cyclists and pedestrians. The advantages of the system are automation of the data collection and filtering, continuous description of road users' trajectories and possibility for making observations over long periods. Special techniques (stereo vision, fitting Euclidean transformations) are tested to further improve the accuracy of the position and speed estimations. (Author/publisher) This publication may be accessed by Internet users at: http://www.ictct.org/workshop.php?workshop_nr=25
Samenvatting