A new method for quarter mirror filter bank design

Abstract

In recent years multirate filter banks, which are part of the system structures in signal processing and communication systems, biomedical signal processing, the sub-bands coding processing are often used. A very important place, especially in the design of filter banks having frequency selectivity, the filter bank is a measure of the performance of different frequency bands. Filter bank must have high stop band attenuation and narrow pass band for good frequency selectivity of the filter bank. In the design and optimization based on filter banks, iterative methods are used. Iterative methods can be exampled such as Hooke and Jeeves algorithm, the Parks-McClellan algorithm, Lagrangian method, Fibonacci equations, Golden search algorithm. This study is based on Parks-McClellan algorithm which is known iterative method, provides desired optimal impulse response with minimal error rate and is one of the best known technique. In this study, a new algorithm which is based on optimizing the cutoff frequency of prototype filter has been developed. Based on windowing technique filter design with the proposed new QMF-Quadrature Mirror Filter bank designed. Designed filter banks effectiveness is tested on two different input signals and showed with figures.