Vehicle Identification Research Based On Feature Fusion Of Acoustic And Seismic Signals

Automatic vehicle identification is an important part of traffic management system, andfast and efficient methods of automatic vehicle identification has an important role inintelligent traffic management. It is hot spots how to use new technology to achieve automaticvehicle recognition in the field of intelligence transportation systems research.As vehicleacoustic and seismic signals have advantages of little information amount and goodenvironment adaptability,vehicle identification based on the acoustic and seismic signals hasbeen widely concerned at home and abroad in recent years.For the defects of the current recognition methods, this paper proposes a vehicleidentification method based on the fusion of acoustic and seismic signals.On the basis ofanalyzing the situation of vehicle recognition with the acoustic and seismic signals, this paperanalyzes the compositions, characteristics,and affecting factors for the signal propagation ofvehicle sound and vibration signals.Then this paper mainly focuses on the commonly usedsignal features and feature extraction methods, proposes a new Block Cepstrum Summationfeature extraction methods combined with cepstral theory, analyzes the principle of fusiontechnology and the advantages and disadvantages of various levels of data, and form a featurevector using feature-level fusion. Finally, support vector machines is used to precess acousticand seismic signals for two different types of vehicles, and experimental results show that theBCS feature extraction algorithm proposed in this paper can effectively extract features ofacoustic and seismic signals,it is a feasible vehicle recognition method based on the fusion ofacoustic and seismic signals, and recognition rate is over85%.