This paper proposes a design method of stack filters using genetic programming (GP). The proposed method formulates the design of stack filters as the desicion problem of positive boolean function (PBF) which minimizes the mean square error between the desired signal and the filter output signal. The desicion problem of PBF is considered as the combination problem of logical AND/OR operations and variables constituting the PBF. GP is applied to search PBF. Moreover, this paper also describes a GP-based design method of stack filters taking symmetry into consideration. Experimental results show that stack filters designed by the proposed method have superior performance in impulsive noise suppression to ones designed by the traditional method.
展开▼