In physical optics prediction of radar cross section, the surface of the object is usually modeled by flat triangular meshes in order to simplify the intersection test and electromag- netic field calculations. This geometrical representation ignores the surface curvature and the resulting changes in the curvature of the reflected wavefront. As a result the accuracy of radar cross section prediction decreases. This paper uses a bilinear approximation method to estimate the normal vector variations of surface and then uses the curvature tensor to recover the surface curvature data from flat meshes. Numerical results show that the presented method properly recovers the surface curvature at a low computational cost. By using the recovered curvature data in physical optics calculations, the accuracy of radar cross section prediction of a missile model is enhanced 1dB on average.

Keywords