SAR Target Detection Based on NCM-1K Clustering

In training-free Synthetic Aperture Radar (SAR) target detection, it is difficult to distinguish targets and non-targets accurately. To solve the problem, a new improved NCM clustering method with one clustering center (K=1), as called NCM-1K clustering method, is proposed in this paper. The principle of NCM-1K is to divide the potential target areas into target areas and non-target areas according to the Polar Scale-Invariant Feature Transform (PSIFT) descriptor. The proposed NCM-1K clustering method can realize the SAR target detection with free-training. Experimental results verify the feasibility and effectiveness of proposed method.