DATA EXTRAPOLATION FOR HIGH-RESOLUTION RADAR IMAGING

In radar imaging, AR modeling is sometimes used to extrapolate the sca ttered field data to obtain a high resolution image. In general, the B urg method is used to estimate the prediction parameters. The Burg met hod leads to a stable prediction biter but can also cause bias in the estimated spectra. One can also use the modified covariance method (MC M) to estimate the prediction parameters. These parameters lead to unb iased spectra. However, the MCM does not guarantee a stable prediction filter. One may have to modify the prediction parameters to ensure a stable prediction filter. One way to ensure stability is to reflect th e unstable poles inside the unit circle. It is shown that the modified parameters can be used effectively for data extrapolation. The radar images obtained using this extrapolated data are more accurate than th ose obtained using the extrapolated data from the Burg prediction para meters.