Abstract: A continuous time-variant digital filtering method and apparatus. In particular, a method for continuous time-variant digital filtering of a digitized signal comprrises the steps of windowing the input digitized signal into discrete windows. For each discrete window of the input digitized signal, an autocorrelation matrix is determined. Thereafter, a filter coefficient matrix for each window of the input digitized signal is determined from the autocorrelation matrix. Corresponding filter coefficients from each filter coefficient matrix and window of the input digitized signals are sequentially collected and interpolated for each sample point of the input digital signal to form continuous filter operators. Each continuous filter operator is sequentially convolved with the input digitized signal to form a partial output signal and the partial output signals from the convolution of each continuous filter operator with the input digital signal are summed to form a final output signal.

Abstract: At least two methods of reconstituting the original form of a time/amplitude trace, such as seismic data, are disclosed wherein the seismic data has been rectified by a scanner to produce a signal representative of a hardcopy display of the trace. One of the methods comprises selecting a time window of a certain duration, obtaining a measurement of the predominant period within the selected positive components of the signal, and filtering the signal with a bandpass filter having a range of about F/2 to about 3F/2, where F equals the reciprocal of the predominant period of the signal.

Abstract: A method and apparatus for digital time-variant filtering are disclosed. The digital time-variant filters are implemented using the Hilbert transform and a time-varying instantaneous frequency control function for controlling the cutoff frequency. The digital time-variant filters are adaptable to various filtering processes, that is, as either a high-cut, low-cut, band-pass, or reject filter. The various digital time-variant filters are stable in that no unusual transients occur in the output signal; are flexible in that the cutoff frequency is controlled by a selectable time-varying control function signal; and are efficient in that computational requirements are only twice to 16 times that of known digital time-invariant filters.