Sectioning techniques for processing signals with signal-dependent noise
B.R. Hunt, H.J. Trussell · 2005
In a previous paper, it was shown that a non-linear Bayesian approach to deconvolution could be implemented by sectioned digital filtering. The method of solution used an iterative scheme to maximize the a-posteriori probability in a Bayesian estimator, and it was demonstrated that the iteration computations could be carried out by using sectional digital filtering. In this paper, it is shown that the same technique can be used to solve a wider class of problems of a nonlinear Bayesian estimation with noise which is dependent upon the signal. The analysis is general enough to encompass signal deconvolution, as well as noise smoothing, in both the linear and nonlinear cases.