Least square filters for approximating perfect reconstruction filter banks
Mark S. Spurbeck, Louis L. Scharf · 2002
This paper presents a least squares (LS) design methodology for approximating perfect reconstruction filter banks. A filter bank can be represented as a multi-input multi-output (MIMO) LTI system whose transfer function is described by the filter bank polyphase matrix. Rising this MIMO system representation, we frame a general least squares filter design problem. Then given an arbitrary set of rational analysis filters we find the causal synthesis filters for a filter bank which achieves the best causal LS approximation to perfect reconstruction.>