作者: Zongjie Cao , Jilan Feng , Rui Min , Yiming Pi
关键词:
摘要: Feature extraction is a very important step in Synthetic Aperture Radar automatic target recognition (SAR ATR). In this paper, feature procedure based on the nonnegative matrix factorization (NMF) and Fisher linear discriminant (FLD) analysis proposed for SAR images. Firstly, segmented images are processed by NMF algorithm, which can extract features that contain local spatial structure information of targets. Then FLD method applied to extracted features, thus discriminability be enhanced. Both locality separability between classes enforced two-phase extracting procedure. Finally, obtained used recognition. Compared several other methods, experimental results show effectiveness