Cumulant-based LP method for two-dimensional spectral estimation

摘要

A cumulant-based linear prediction (CBLP) method fortwo-dimensional (2-D) spectral estimation is presented. The main idea ofthe method is to compute the coefficients of two differentsingle-quadrant prediction filters by applying the LP theory to aselected 2-D fourth-order mixed cumulant slice of the noisy signal.These coefficients are employed in formulating two differentautoregressive spectral models. Both spectral models are combined toobtain the desired spectral estimate. The effectiveness of the proposedCBLP method is demonstrated through computer simulation