A NOVEL KNN CLASSIFIER FOR ACOUSTIC VEHICLE CLASSIFICATION BASED ON ALPHA-STABLE STATISTICAL MODELING

This paper describes a novel methodology for statistical modeling and classification of acoustic signals collected from a wireless sensor network. Our S alpha S kNN classifier is based on a variation of k-nearest neighbor algorithm. First, we perform a 1-D wavelet decomposition of the acoustic signal and we model the resulting subband coefficients using the alpha-stable distribution. Subsequently, the alpha-stable distribution parameters are estimated during feature extraction and the similarity between two acoustic signals is measured by employing a variant of the Kullback-Leibler Divergence (KLD) between the characteristic functions of the corresponding subband representations. We evaluate and compare the performance of the proposed methodology by using actual recorded data in a vehicle classification application.