EM Algorithms for Parameter Estimation in Non-Gaussian Noise

EM algorithm,which uses incomplete data in solving the model parameters of maximum likelihood estimation,is a fairly good algorithm for non-Gaussian noise parameter estimation problem.Non-Gaussian noise parameter estimation is the main difficulty,when sufficient statistic does not exist,This means that the estimation from the observation of space down to the mapping of space relies on the attempted estimation parameters.When the noise probability density is unknown,EM algorithm could accurately estimate the parameters of non-Gaussian noise,and the estimation variance is close to the C-R lower bounds.