Synthetic aperture radar target configuration recognition using locality-preserving property and the Gamma distribution

摘要

Taking the inevitable speckle noise in synthetic aperture radar (SAR) images into account, a feature extraction algorithm is proposed by combining the locality-preserving property and the Gamma distribution for SAR target configuration recognition in this study. To describe the essential characteristics of SAR images, the authors model the speckle noise by the Gamma distribution. Moreover, the locality-preserving property which is of great importance for recognition is fused into the statistical model to preserve the local structure of the datasets. Besides, they modify the affinity matrix that is used in the locality-preserving property to capture both the local and global structures of the datasets for configuration recognition. The coordinate descent method is carried out for parameter estimation, in which a decomposition method is presented to obtain the transformation matrix. The experimental results on the moving and stationary target acquisition and recognition database validate the effectiveness of the proposed algorithm.