System for vehicle classification: Hardware prototype and off-line signal proccesing

This paper presents a complete system for vehicle classification composed by an inductive-loop detector and the corresponding off-line algorithms. The system detects the presence of vehicles by means of a shift in the oscillation period of the selected loop so that the signature of the detected vehicles is registered by measuring the duration of a fixed number of oscillator pulses. We focus on the open issue of counting the number of vehicles (classified into cars, vans and trucks) on a roadway. The classical method for such purpose consists of estimating the vehicle length using the inductive signatures obtained from two loops and, subsequently, it classifies them taking into account a prefixed threshold. With the goal of improving the performance, in this paper we propose a method based on classifying the vehicles with uncertainty about the decision. Such methods are evaluated using real inductive signatures captured with our prototype.