Vision based vehicle classification offers substantial advantages compared with traditional vehicle classification systems, which use information from inductive loops, axle counters, Doppler radar and infra red light curtains to classify vehicles. The principal benefits include the ease of installation and maintenance, as well as much greater flexibility, performance and cost effectiveness. This paper details a vision based classification system, which uses image processing and neural network techniques to classify vehicles, looks at the performance of the system, and discusses problems associated with vision systems. (Author/publisher) For the covering entry of this conference, please see ITRD abstract no. E209471.
Abstract