Performance of Blackman window family in M-channel Cosine Modulated Filter Bank for ECG signals

Abstract

In this work, 32 channel Cosine Modulated Filter Bank (CMFB) is used to decompose and reconstruct single lead ECG signal. The prototype filter for analysis and synthesis filter banks is designed using window technique. Variants of Blackman window, which is a fixed window, are used to design the prototype filter. The cutoff frequency of prototype filter is optimized to minimize the error function. A linear iterative optimization algorithm is used to achieve minimum reconstruction error. The results show a significant reduction in reconstruction error. In some cases, Blackman window family performed even better than adjustable windows. Further, No perceptible loss in signal morphology and features are visible in reconstructed ECG signal, which is in accordance with clinical requirement. ECG signals from MIT-BIH ECG database is used to test the designed filter bank.