Vehicle Speed Recognition from Noise Spectral Patterns

The use of noise maps in large cities has become a common practice since they have been recognized as a powerful tool for determining the population exposure to environmental noise and, therefore, a valid aid to identify the most appropriate mitigation actions. The possibility to complement noise maps with further information such as the vehicle speed distribution, can represent an additional benefit to local Municipalities. The well-known fact that speed is the most relevant characteristic to discriminate between different vehicle noise spectra, led us to perform a dedicated measuring campaign devoted to record both vehicle features (noise spectrum and speed). The recorded spectra have been statistically analyzed and classified according to their speeds. Three main spectral patterns, corresponding to different mean speed, have been derived. Such patterns have been used, first, for discriminating among non-vehicle noises and, second, they have been associated with vehicles travelling within a certain speed interval. The results suggest that this method can prove useful in practical situations in which traffic noise and vehicle mobility need to be controlled/assessed