Target Recognition in Ultra-Wideband SAR Imagery

摘要

Ultra-wideband (UWB) synthetic-aperture radar (SAR) images-with greater than 95-percent bandwidth occupancy-provide the potential for recognition of targets embedded in foliage; recognition methods in these applications are based on wideband resonant signatures of the targets. Currently, resonance-extraction techniques hinge on contemporary adaptations of Prony's method; this method, however, has poor performance in the presence of noise and is very computationally intensive. A form of resonance analysis is proposed that applies linear-transform methods. Both the Fourier basis and two multiresolution bases-the Haar wavelet and the Gaussian basis-were employed in the analysis. Target declaration confidences were established by simple correlation of the two sets of spectral coefficients-one set from the transformed data, and the other from a synthetic template generated from both prediction and empirical observation. This permits a fast, efficient scheme for recognition of target resonance effects in wideband imagery. Five UWB images from the Army Research Laboratory's UWB SAR instrumentation system were analyzed through the use of canonical targets (dipoles) of differing dimension and orientation. Results are presented and summarized for each of the targets and transform methods employed in the analysis. Ultra-wideband, Synthetic aperture radar, SAR, Automatic target recognition, ATR.