Classifying buried and obscured targets of interest from other natural andmanmade clutter objects in the scene is an important problem for the U.S. Army.Targets of interest are often represented by signals captured usinglow-frequency (UHF to L-band) ultra-wideband (UWB) synthetic aperture radar(SAR) technology. This technology has been used in various applications,including ground penetration and sensing-through-the-wall. However, thetechnology still faces a significant issues regarding low-resolution SARimagery in this particular frequency band, low radar cross sections (RCS),small objects compared to radar signal wavelengths, and heavy interference. Theclassification problem has been firstly, and partially, addressed by sparserepresentation-based classification (SRC) method which can extract noise fromsignals and exploit the cross-channel information. Despite providing potentialresults, SRC-related methods have drawbacks in representing nonlinear relationsand dealing with larger training sets. In this paper, we propose a SimultaneousDecomposition and Classification Network (SDCN) to alleviate noise inferencesand enhance classification accuracy. The network contains two jointly trainedsub-networks: the decomposition sub-network handles denoising, while theclassification sub-network discriminates targets from confusers. Experimentalresults show significant improvements over a network without decomposition andSRC-related methods.
展开▼
