SαS Process Location Parameter Adaptive Estimator Based on Data Censoring

The task of location parameter estimation by means of adaptive censored procedure in SαS noise environment is considered. Based on the carried out research it is proposed to use the percentile coefficient of kurtosis and median absolute deviation as parameters able to describe processes with SαS pdf in case of absence of a priori knowledge about its statistical characteristics. The new formula is presented to evaluate the censoring parameter of the investigated estimator. Based on comparative analysis it is shown that the proposed adaptive procedure allows one to obtain the smallest variance among the considered robust and adaptive estimators for α ≪ 0.9. The proposed adaptive estimate can be put into basis of noise suppressing filter to be applied in image homogeneous regions in case of mixed noise.