Self-speed and headway measurement in highway traffic from onboard video footage

Abstract

We present a method to automatically collect data about distance and speed of vehicles ahead of a car participating in highway traffic. The method should support the registration of traffic in front of a driving car over long periods of time in order to be able to detect over-persistent congestion (OPC) situations. The method requires no specialized devices; it is based on a simple video stream from any (low-cost) camera mounted near the front window of a car. The data is used to fine-tune traffic flow models in order to determine the conditions for the emergence and the dissipation of traffic jams more accurately. The paper deals with the performance of the vehicle tracking method itself and presents a criterion for over-persistent traffic jams applied to the data. The outlook towards a driver’s advisory system to reduce traffic congestion (and thus time loss and environmental load) is discussed as well.