A Novel Knowledge-Aided Training Samples Selection Method for Terrain Clutter Suppression in Hybrid Baseline Radar Systems

For a space-based radar system with hybrid baseline, the problem of clutter angle-Doppler spectral broadening poses a significant challenge to clutter cancellation in the terrain fluctuant observation scene. To improve the robustness of clutter suppression, this paper proposed a homogeneous sample selection method based on a novel concept of Generalized Spatial Spectrum Density Function (GSSDF). This method can be summarized as three crucial steps. First, the GSSDF was constructed by the prior information of DEM data, radar system parameters, and the backscattering model. Then, the angle of deflection (AOD) and the equivalent bandwidth (EBW) of GSSDF were adopted to measure the diffusion of clutter spectral. Subsequently, the sample selection criterion was established by a new threshold detection strategy. But the key here is that an appropriate detection threshold of the AOD and EBW can be determined by the characteristic of filter response. To summarize, this approach ensures that the training samples sharing similar clutter properties can be selected to estimate clutter covariance matrix (CCM); thereby, enhances clutter suppression capability. Finally, the experimental results demonstrated that the proposed method can obtain better clutter suppression performance than other contrast methods.