Target Recognition of SAR Image Via Joint Weighted Fusion of Multiple Features

A synthetic aperture radar (SAR) image target recognition method based on multi-feature joint weighted fusion is proposed. Elliptic Fourier Descriptors (EFDs) of the target contour and Non-Negative Matrix Factorization (NMF) of the image are extracted to describe the target geometric shape and the global grey level distribution of the SAR image, respectively. The azimuth sensitivity image is constructed to describe the electromagnetic scattering characteristics of the target. The joint sparse representation is used to make a weighted fusion decision for the three features, so as to give full play to their complementary advantages. Finally, the class of the test sample is determined based on the reconstruction errors from the joint sparse representation. Based on the Moving and Stationary Target Acquisition and Recognition (MSTAR) dataset, experiments are carried out under various operating conditions, which fully verify the effectiveness of the proposed method.