This paper describes the application of the ASSET-2 (A Scene Segmenter Establishing Tracking, Version 2.0) image processing system, developed at DERA Chertsey, to the counting and tracking of pedestrians and vehicles. Post-processing is applied to the output of ASSET-2, which segments video images into separately moving objects on the basis of image motion, to discriminate between vehicles and pedestrians and to derive object counts and tracks across designated virtual trip-wires within the scene. This research has been undertaken in support of Pedestrian Behaviour and Exposure Studies, being carried out by Ross Silcock Limited and funded by the U.K. Department of the Environment, Transport and the Regions. The purpose of these studies is to quantify pedestrian exposure to risk in crossing urban roads through the analysis of pedestrian and driver behaviour to establish the requirements for longer term monitoring; to examine priorities between behaviour modification and engineering measures; and to provide policy development guidance. (A)
Abstract