FFT-Accelerated 3-D Near-Field Radar Image Generation for Objects Over Rough Surfaces

Conventional three-dimensional (3-D) near-field radar image simulation requires electromagnetic (EM) scattering data collection over a densely sampled frequency-azimuth-elevation sector, resulting in burdensome or even impractical computational complexity. In this paper, a fast Fourier transform (FFT)-accelerated 3-D near-field radar scattering image generation technique is proposed for objects over rough surfaces, applicable at arbitrary observation angles. A closed-form expression for the 3-D spatially variant point spread function (SVPSF) is derived to directly calculate near-field image-domain contributions from ray tubes traced using the multipath model enhanced shooting and bouncing rays (MP-SBR) [30]. The multiple scattering effect is addressed by developing a spatial mapping process. The 3-D radar image formation is formulated as a convolution integral, allowing the FFT-acceleration scheme to significantly reduce the computational complexity. Moreover, the 3-D resolution equations are derived and used to evaluate radar image quality. Simulated radar images are compared with both measured and simulated counterparts by conventional technique to demonstrate the high fidelity and efficiency of the proposed technique.