Improving similarity measures for re-identification of vehicles using AMR sensors

The use of Anisotropic Magneto-Resistive (AMR) sensors for road traffic monitoring has gained popularity in recent years. The Earth magnetic field variations caused by vehicles passing over this kind of sensors can be measured and used for various purposes, including vehicle re-identification at different sensing zones. AMR sensors provide data that allow the calculation of traffic parameters by using simple signal processing techniques. But when vehicle re-identification is the objective, it is not clear which is the best method to compare the signals from different sensors in order to know whether or not they belong to the same vehicle. This paper presents an experimental study with the objective of determining the optimal similarity measure function for AMR sensors signals when used for vehicle re-identification.