Design of Acoustic Vehicle Detector with Steady-Noise Suppression

Vehicle detection is a basic component for many applications in intelligent transportation system (ITS). We are developing a low-cost vehicle detector relying on sound arrival time difference on two microphones. Our previous paper presented that our acoustic vehicle detector successfully detected vehicles with an F-measure of 83 %. However, the acoustic detector has difficulties in vehicle detection in an environment with steady noise such as rain noise.This paper presents a steady-noise suppression method for the acoustic vehicle detector. Our key idea is to exclude the influence of steady noise in a sound delay estimation. The acoustic vehicle detector estimates vehicle sound delay by finding a peak on a cross-correlation function. We theoretically analyze the influence of steady noise and remove a peak caused by the noise to minimize the influence. Experimental evaluations revealed that the steady-noise suppression method effectively reduced the noise influence and resulted in F-measures of 0.92 and 0.90 in normal and heavy rain conditions, respectively.