Dimensionality reduction for acoustic vehicle classification with spectral clustering
Classification of vehicles has broad applications, ranging from traffic flow management to military target recognition. We propose a method for automated identification of moving vehicles from roadside audio sensors. We use a short-time Fourier transform to decompose audio signals, and treat the frequency signature at each time window as an individual data point to be classified. Using spectral clustering, we then decrease the dimensionality of the data sufficiently for K-nearest neighbors to provide accurate vehicle identification.
READ FULL TEXT