This paper presents the results of a study of automatic vehicle classification equipment for high speed toll stations. The aim of the study was firstly to determine the accuracy limits of the HESTIA traffic analysis station. The study then investigated whether it was possible to increase accuracy of the system by using 3 consecutive stations laid out over a distance of 60 metres. The system had to recognize 34 different vehicle categories, from motorcycles to seven axle heavy goods vehicles. The accuracy check was carried out by comparing the vehicle by vehicle listing provided by the system with a time and dated video recording of the traffic. The analysis was extremely lengthy. The results indicate that the error rate for vehicle detection was 0.1% and 0.05% for toll category classification.
Abstract