Removal of Intervening System Distortion by Deconvolution

A numerical procedure (called deconvolution, inverse convolution) is presented for the removal of measurement distortions introduced by intervening systems whose input and output are connected by convolution. The system impulse response is assumed known. Rather than modify this impulse response to avoid calculating unreasonable inputs as is commonly done, the technique developed makes corrections to the given (usually noisy) measured output arrays. The correction is determined by requiring the calculated input to have minimal energy. The mathematical manipulations can be carried out entirely in the time domain, the greatest computational difficulty being matrix inversion. The features of the procedure are conceptual simplicity and the need for only rudimentary subjective judgments from the investigator.