Adaptive method for 1-D signal processing based on nonlinear filter bank and Z-parameter

An approach to synthesis of adaptive 1-D filters based on nonlinear filter bank and the use of Z-parameter is put forward. The nonlinearity of elementary filters ensures the predetermined robustness of adaptive procedure with respect to impulsive noise and outliers. In turn, a local adaptation principle enables to minimize the total output error being a sum of the residual fluctuation component and dynamic errors. The use of Zparameter as a local activity indicator permits to “recognize ” the signal and noise properties for given fragment quite surely and, thus, to select a proper filter from the bank at disposal. Numerical simulation results confirming the efficiency of the proposed approach are presented.