New Design Methodology of an Advanced Optimal Space/Spatial-Frequency Filter

Abstract

New methodology in the creation of the region of support of the optimal (Wiener) filter used in the estimation of two-dimensional (2D) nonstationary signals is considered. The methodology is based on the symmetrical spreading of the filter's region of support around the detected local frequency (LF) of the estimated 2D signal and on simultaneous inclusion (in the region of support creation) of points having equal distance from the detected LF. The proposed filter provides signal reconstruction free of distortion of the original (noiseless) signal.