|
|
A Novel Feature Extraction Method for Unexploded Ordnance in Ultra-Wideband SAR |
Jin Tian; Zhou Zhi-min |
College of Electronic Science and Engineering, National University of Defense Technology, Changsha 410073, China |
|
|
Abstract Ultra-WideBand Synthetic Aperture Radar (UWB SAR) can detect shallow buried UneXploded Ordnance (UXO), where the efficiency of extracted features determines the detection performance. The traditional subband-subaperture processing extracts the frequency and aspect information of UXO scattering with sacrifice of spatial resolutions. According to the issue, in this paper, the frequency and aspect angle features extraction method is proposed on the Space-Wavenumber Distribution (SWD), which is combined with the moment invariants to obtain the shift- and rotation-invariant feature set. The proposed feature set contains not only the frequency and aspect information but also high spatial resolutions information. Field data processing results show that the proposed features in this paper can improve the UXO detection performance efficiently.
|
Received: 02 July 2007
|
|
|
|
|
|
|
|