Study and Performance Evaluation of Two Iterative Frequency-Domain Deconvolution Techniques

Abstract

The performance of two iterative frequency-domain deconvolution techniques, the optimum compensation and the Guillaume-Nahman, is evaluated. The study involved the characterization of the adaptive filters utilized to reduce the deconvolution noise. Comparisons between the two techniques are performed for various classes of signals having different levels of acquisition noise. It is found that the Guillaume-Nahman technique is potentially more accurate but more sensitive to acquisition noise. It is also found that the optimum compensation deconvolution technique uses an adaptive filtet with the passbands coinciding with the frequency bands of the signal, thus optimizing noise filtration, while the Guillaume-Nahman technique utilizes a low-pass filter for all signals.