Traffic Analysis in Roundabout Intersections by Image Processing

Abstract In this paper a method for the evaluation of roundabout performance, based on the image processing of field survey data, is presented. The methodology of investigation consists of the following three main steps: a field survey to collect vehicular flow images using video cameras, the processing of images gathered by the VeTRA (Vehicle Tracking for Roundabout Analysis) software program and the analysis of the trajectories extracted. The methodology enables the automatic computation of the information required to rank and evaluate a generic roundabout: the Entry/Exit (E/E) matrix with vehicle classification (i.e. heavy, light and motorbikes), identification of vehicle trajectories, and extraction of vehicular speed/curvature diagrams along the paths inside the roundabout. The proposed image processing stage overcomes classic problems encountered with image processing such as shadows, object occlusions, and the unpredictable influence of wind or clouds. Calibration and error evaluation are obtained from data collected by a high precision GPS-RTK system mounted on a probe vehicle.