Road traffic monitoring using image processing

A method is described which provides data on traffic congestion over a length of roadway using inexpensive image processing and computing hardware. It provides distinct advantages over more conventional methods since data is collected over a region rather than a point in the road network. Furthermore, the state of the network can be deduced within a time scale comparable to the period with which congestion occurs on a busy road after an incident has occurred. Action can be taken on information acquired directly from the scene and not inferred from statistics built up over a period of minutes. Further work is required to take account of minor perspective distortion effects, which can cause false alarms when single vehicles occupy the distant end of the window. A measure of the amount of traffic in the window will also be needed to enable comparison with other regions of the network. Night time scenes have not been included in the test program. >