LATTICE-STRUCTURED SCHUR-TYPE ELIMINATION OF NARROW-BAND BACKGROUNDDISTURBANCES FROM ACOUSTIC SIGNALS OF VEHICLES IN MOTION

摘要

The paper deals with the problem of detection of vehicles inrnmotion, using emitted acoustic signals. Due to the presencernof background disturbances, the registered real-life signalsrncontain a mixture of a source signal (emitted by the vehicle),rnand a corrupting signal (narrow-band “background noise”rnof bandwidth similar to the original signal). A new methodrnof the background noise elimination is proposed in this paper.rnThe method is based on the lattice-structured Schurtypernorthogonalization/parametrization algorithms in whichrnthe so-called backward estimation errors, computed for thernincoming signal, constitute an orthonormal basis. The proposedrnsolution employs the orthogonal joint-process estimationrnfilter which decomposes the incoming signal using anrnorthonormal basis, derived exclusively for the backgroundrnnoise. The effectiveness of the method is discussed as wellrnas the noise cancellation results, also using real-life signals,rnare presented.