The counting and classification of vehicles is an important partof transportation engineering. In the past 20 years many automated systems have been developed to accomplish that labor-intensive task.Unfortunately, most of those systems are characterized by inaccurate detection systems and/or classification methods that result in many classification errors, thus limiting the accuracy of the system. This report describes the development of a new vehicle classificationdatabase and computer program, originally designed for use in the two-point-time-ratio method of vehicle classification, which greatly improves the accuracy of automated classification systems. The program utilizes information provided by either vehicle detection sensorsor the program user to determine the velocity, number of axles, andaxle spacings of a passing vehicle. It then matches the axle numbers and spacings with one of thirty-one possible vehicle classifications and prints the vehicle class, speed, and wheelbase lengths. It also tabulates and prints totals and average speeds for each vehicle type. This paper describes the database built and utilized, as well as a roadside experiment conducted to test the accuracy of the database and the classification program, showing the classifications to behighly accurate. This paper appears in transportation research record no. 1215, Pavement management and rehabilitation.
Abstract