Multiresolution feature extraction in target detection of UWB SAR

This paper firstly analyzes the difference between high-frequency synthetic aperture radar (SAR) and low-frequency ultra-wide band synthetic aperture radar (UWB SAR) in extraction of multiresolution feature. And then we establish the equivalent models of target and trunk clutter in UWB SAR images and analyze the differences of multiresolution characteristic between target and trunk clutter as image resolution is varied from fine to course. Two forms of first-order autoregression (AR) model are introduced to model multiresolution sequences, and a two-dimensional (2-D) definition of generalized likelihood ratio (GLR) is presented to improve the robustness of multiresolution feature extraction. At last a feature extraction experiment of multiresolution based on region of interest (ROT) is accomplished. The results of experiment testify that the multiresolution feature is effective to improve the performance of UWB SAR target detection.