Fast implementation technique of adaptive Kalman filtering deconvolution via dyadic wavelet transform

A new approach of adaptive Kalman filtering deconvolution (AKFD) via dyadic wavelet transform is proposed. To the computing complexity of the approach , a fast implementation technique is proposed. The AKFD via dyadic wavelet transform discards the assumption of stationarity of signals in predictive deconvolution, and solves the problem of improving resolution at the price of decreasing signal-to-noise rate (SNR) obviously, consequently it has a b etter ability of resistance noise. Suppressing false reflections and improv ing resolution in dyadic wavelet transform domain is better than that in ti me domain. At the same time, the approach also overcomes the drawback of in creasing the low-frequency component of AKFD in time domain. The fast imple mentation technique makes use of the assumption of local stationary for 2D seismic data. The technique reduces the calculation amount of the adaptive predictive operators by calculating an adaptive predictive operator in each segment, and then applying the algorithm of spline interpolation to interp olate in transverse and in portrait. A great deal of experiments indicates that the computing speed can be increased hundreds times, and the original calculation effect is still maintained.