The use of an ultra-wideband (UWB) synthetic aperture radar (SAR) for detection of buried or obscured targets is an intense area of research because of the penetration and high-resolution capabilities of the radar. We address the important question of clutter modeling. More specifically, we examine the clutter observed from a forest for robust detection of targets that are buried in a forest environment and obscured by foliage. The observed clutter is a highly impulsive random process that can be accurately modeled with the previously proposed class of alpha-stable processes. Although these /spl alpha/-stable distributions also accurately model the behavior of impulsive sonar sea clutter and impulsive noise in a communication channel, we examined only the performance of a receiver for detecting targets in radar clutter based on alpha-stable distributions via extensive testing on real data, and we compared the performance of receivers based on more popular clutter distributions, such as the K and the Gaussian. The results show that, at sufficiently low probability of false alarm rates, the receiver based on the alpha-stable distribution outperformed the K- and Gaussian-based receivers.
展开▼
