Construction of 2D training database based on high‐frequency RCS prediction methods

Abstract

AbstractFor automatic target recognition using inverse synthetic aperture radar images, training data used have been mostly obtained from measured targets.However, because of cost and secrecy, securing an actual enemy aircraft is very difficult. In this article, to determine whether this problem can be solved using computer simulation, we constructed a training database using three high‐frequency radar cross‐section prediction methods: physical optics (PO), physical theory of diffraction (PTD), and shooting and bouncing ray (SBR). Then, we used the polar mapping classifier to quantify their accuracy in classifying images. The combinations of SBR with PTD and PO with PTD had almost the same accuracy and both were more accurate than when SBR and PO were used alone. © 2010 Wiley Periodicals, Inc. Microwave Opt Technol Lett 53:223–229, 2011; View this article online at wileyonlinelibrary.com. DOI 10.1002/mop.25652