A Training Sample Selection Method With Fusing GIP Statistic and Geographic Information for Airborne Radar

ABSTRACT A training sample selection method is proposed by fusing the generalized inner product (GIP) statistic with geographic information to construct the fused metric of the sample set. Based on this metric, more suitable training samples are selected. The experimental results demonstrate that the proposed method exhibits excellent robustness in different clutter environments, particularly in complex ground environments containing discrete moving scatterers, where its clutter suppression performance is superior to that of both GIP and knowledge‐aided space‐time adaptive processing (KA‐STAP) methods.