High resolution radar imaging using bandwidth extrapolation

The resolution of a synthetic aperture radar (SAR) image, in range and azimuth, is limited by the transmitted signal bandwidth and the synthetic aperture length, respectively. Bandwidth extrapolation algorithm is one of the various superresolution techniques for improving the resolution. The frequency AR model is used in the paper to expand the effective bandwidth of the signal based on the bandwidth extrapolation together with linear prediction algorithm. The performance estimation function, namely the prediction error as a scaling power-law function of the linear error prediction filter length, is derived through analyzing large numbers of computer simulation experiments. It is of considerable importance to determine the linear error prediction filter length for optimizing the bandwidth extrapolation parameters. It is demonstrated by the theoretical analysis and simulation results that the bandwidth extrapolation method is feasible and effective for expanding the bandwidth to obtain higher resolution SAR image by applying the method to the range and azimuth directions, respectively.