Air-or vehicle-borne Ultra-Wide Band Synthetic Aperture Radar (UWB SAR) has the capability to detect landmines and landmine fields over large area from a standoff distance quickly

which is the trend of landmine detection.In order to reduce the false alarms in landmine detection

it is needed to extract efficient landmine feature to classify landmine and clutter.In this paper

a High-Dimensional Time-Frequency Distribution (HDTFD) based method to extract the aspect invariance feature of landmine scattering is proposed

where the realization can be performed with High-Dimensional Wigner-Ville Distribution (HDWVD) and High-Dimensional Choi-Williams Distribution (HDCWD).The proposed HDTFD based feature extraction method can extract the frequency and aspect angle dependent information in scattering function efficiently while maintains high spatial resolutions.The HDWVD and HDCWD based feature extraction methods have been compared using the real data collected by the Rail-GPSAR system in terms in resolution

cross-terms suppression and aspect invariance feature extraction.Field data processing results have shown that the HDCWD based feature extraction method is more suitable for real data processing.