Results on AR-modelling of nonstationary signals
Signal Processing
Neural Networks: A Comprehensive Foundation
Neural Networks: A Comprehensive Foundation
Digital Image Processing
IEEE Transactions on Signal Processing
An adaptive optimal-kernel time-frequency representation
IEEE Transactions on Signal Processing
Generalized feature extraction for time-varying autoregressivemodels
IEEE Transactions on Signal Processing
A Wavelet Packet Algorithm for Classification and Detectionof Moving Vehicles
Multidimensional Systems and Signal Processing
Vehicle classification in distributed sensor networks
Journal of Parallel and Distributed Computing
Wavelet-based acoustic detection of moving vehicles
Multidimensional Systems and Signal Processing
A diffusion framework for detection of moving vehicles
Digital Signal Processing
Vehicle classification in wireless sensor networks based on rough neural network
ISNN'06 Proceedings of the Third international conference on Advances in Neural Networks - Volume Part III
Hi-index | 0.00 |
Time-varyingautoregressive (TVAR) modeling approach for the analysis of acousticsignatures from moving vehicles is presented in this paper. Acousticsignatures from moving vehicles are nonstationary, and featuresextracted under the stationary assumption often result unsatisfactoryperformance. In TVAR modeling approach, the time-varying parametersare expanded as a linear combination of deterministic time functions.In this paper, the TVAR parameters are expanded by a low-orderdiscrete cosine transform (DCT), since DCT is known to be closeto the optimal Kahrunen-Loève transform when the signalis Markov. The maximum likelihood estimation and order selectionin TVAR models are also discussed. Many attributes of vehicleactivities, such as vehicle type, engine speed, loading, roadcondition, etc., may be inferred from the estimated model parameters.The performance of the TVAR modeling approach is tested withboth synthetic and real acoustic signatures. A synthetic signalcontaining multiple time-varying sinusoids are used to comparethe performances in the estimation of time-frequency distributionwith other approaches. In the experiment with acoustic signaturesfrom moving vehicles, it is shown that the TVAR models can beeffectively used to determine vehicle activities and types atclose range and cruising speed.