A Bayesian approach to harmonic retrieval with clipped data

In this paper, harmonic retrieval is addressed under the standard assumptio n of observations corrupted by an additive white Gaussian noise but also in the presence of hard clipped observations. A Bayesian approach to solve th ese problems is proposed. Bayesian models are first presented that allow us to define posterior distributions on the parameter space. All Bayesian inf erence is then based on these distributions. Unfortunately a direct estimat ion of these distributions and of their features requires evaluation of som e complicated high-dimensional integrals. Efficient stochastic algorithms b ased on Markov chain Monte Carlo methods are presented here to perform Baye sian computation. In simulation on synthetic and real data sets, these algo rithms allow the estimation of the unknown parameters in difficult conditio ns. (C) 1999 Elsevier Science B.V. All rights reserved.