M-channel Sub-band Noise Signal Reconstruction

MULIRATE analysis/synthesis system based on filter banks are widely used for time-frequency decomposition (analysis) and reconstruction (synthesis) in many applications areas.In this paper,a multirate Kalman sysnthesis filter is introduced to replace the ocnventional synthesis filter to achieve optimal reconstruction of the input signal with the consideration of subband noises.At first, the concept of perfect reconstrution of input signal from the view of filter banks and wavelet transform are introduced,.and then the analysis/synthesis system is analyzed in time domain ,third ,the kalman filter is explained,finally ,a multirate kalman synthesis filter and multirate kalman filterirng algorithm are intruduced to achieve optimal reconstruction of the input signal with the consideraion of the subband noises.Simulation results show that the performance improvement of signal reconstruction in noisy filter bank system is remarkble.