Acoustic information fusion for vehicles identification

In the article an information fusion approach for vehicles classification using acoustic signal was presented. Many acoustic features can contribute to a right diagnosis of the vehicle. Consisting only one set of features can omit the relevant information. It is possible to improve the accuracy of classification thanks to the technique of information fusion, where various aspects of acoustic `fingerprint´ are being taken into consideration. Two sets of features of the signal were distinguished: based on frequency analysis (harmonic elements) and based on multidimensional correlation relations. Using SVM and k-NN classifiers identification of the given class of vehicles is being made. A vehicle different classes audio signal database was used for the assessment of effectiveness of the proposed solution. Results are showing the improvement the effectiveness of recognizing towards applying only of one features set of the vehicle.