Matching-Pursuits Based Feature Extraction with Reduced Aspect Sensitivity for Ultra Wide-band Radar Target Identification

摘要

This paper presents a matching-pursuits based technique to radar target feature extraction for aspect sensitivity reduce of template, including a training algorithm of the feature atom dictionary to characterize the target scattering. There is an important contribution in our technique that finite basis atoms, as features, are used to represent the complicatedly scattering behaverior of the whole scattering waveform without any prior parameterization model hypothesis. Furthermore, as demonstrated in this paper, the proposed technique can reduce effectively aspect dependence of the extracted extraction for wide-angle target identification. Synthesized scattered responses and measured scattering signatures in a chamber are used to demonstrate the effectiveness of dictionary feature proposed in this paper.