Scatterer‐based approach to evaluate similarity between 3D em‐model and 2D SAR data for ATR

Abstract

The similarity evaluation between three-dimensional (3D) electromagnetic model (em-model) and 2D synthetic aperture radar (SAR) data is a key factor in em-model-based SAR automatic target recognition (ATR). In this study, a scatterer-based approach is proposed to evaluate the similarity between 3D em-model and 2D SAR data for ATR purpose. A target is characterised by a set of scatterers and the similarity between 3D em-model and 2D SAR data is evaluated through all these scatterers. First, information of each scatterer is predicted from 3D em-model and used to guide the scatterer extraction from 2D SAR data. Then similarity of each scatterer pair is evaluated through a hypothesis testing approach. In the end, these similarities are synthesized based on Dempster–Shafer (D-S) evidence theory as a whole similarity. The innovative contributions of this study are as follows: a scatterer-based similarity evaluation method between 3D em-model and 2D SAR data at arbitrary target pose is established for ATR purpose. This method is able to resist noises and partial occlusion. Besides, from the result, one can attribute physical information to the measured target. Experiments using data simulated by a high-frequency electromagnetic code verify the validity of the method.