Computer-Simulated UWB VHF SAR Targets Based on Statistics of Real Data

Abstract

In this paper, we performed a statistical study of the targets of wavelength-resolution synthetic aperture radar (SAR) images and addressed the problem of limited training samples. Based on suitable statistical models for the real targets of the evaluated SAR image, we considered a sample generator method to create computer-simulated targets. For such, well-known statistical models, namely, Burr, Gamma, Gaussian, Log-normal, Rayleigh, Rician, and Weibull, are considered for the Anderson-Darling (AD) goodness-of-fit test. This test is a statistical nonparametric hypothesis test generally used to investigate whether a given probability distribution yields a good fit for a given data set. The images used in this study were obtained from the Swedish ultrawideband (UWB) very-high-frequency (VHF) SAR CARABAS II system whose consists of 24 magnitude single-look SAR images. The obtained computer-simulated targets followed similar statistical characteristics when compared to the real ones.