Root contours of two-dimensional prediction polynomials

Two-dimensional signals with linear move-out appear in many array signal processing applications such as wide-band direction finding, ground clutter in airborne radar systems, seismic data and so on. We present a simple approach to modeling 2-D signals with linear move-out characteristics. Linear prediction is used to examine the various regions of support for the all-pole modeling of these signals. In addition, the relationship between the slope of the signal and the model parameters is derived and extensions to other regions of support, such as elliptical are presented.