A maximum a posteriori algorithm for reconstruction of targets in incompletely defined correlated noise

The maximum a posteriori (MAP) line spectral estimator used to characterize sinusoids in data corrupted by Gaussian noise of unknown correlation is generalized to the case where an experimental estimate of noise covariance is available. The estimator is robust to noise with mean square error and standard deviation falling below that of the classical MAP for increasing number of samples, while approaching classical MAP for the case of no prior knowledge.