Acoustic information fusion for ground vehicle classification

Many acoustic factors can contribute to the classification accuracy of ground vehicles. Classification based on a single feature set may lose some useful information. To obtain more complete knowledge regarding vehicles’ acoustic characteristics, we propose a fusion approach to combine two sets of features, in which various aspects of an acoustic signature are emphasized individually. The first set of features consists of a number of harmonic components, mainly characterizing engine noise. The second set of features is a group of key frequency components, designated to reflect other minor but also important acoustic factors, such as tire friction noise. To find these features, we apply a harmonic extraction and a mutual information based method that have been shown effective in our previous research. Fusing these two sets of features provides a more complete description of vehicles’ acoustic signatures, and reduces the limitation of relying one particular feature set. Further to a feature level fusion method, we propose a modified Bayesian based fusion method to take advantage of matching each specific feature set with its favored classifier. To assess the proposed fusion method, experiments are carried out based on a multi-category vehicles acoustic data set. Results indicate that the fusion approach can effectively increase the classification accuracy compared to those using each individual set of features. Bayesian based decision level fusion is found to be significantly better than the feature level fusion approach.