A deconvolution approach to moving sources localization
H. Chuberre, J.-J. Fuchs · 2002
The paper addresses the problem of the estimation of moving sources parameters. These sources are supposed to be located in the far-field and their number is unknown. A simplified model of the source motion is used and a deconvolution approach is applied to the output of a conventional beamformer (CBF) that does not assume that the sources are moving. The procedure fits a set of kernels parametrized by bearing, power and motion to the outputs of the CBF over selected spatial sectors and can be seen as an approximate maximum likelihood method since the criterion that is minimized is based on the statistical properties of the beams. Results of simulations are presented which show the potential of this method.>