Motion feature extraction from inverse synthetic aperture radar image time-series: a sparse and low-rank approach

Motion information of the non-cooperative targets observed by inverse synthetic aperture radar (ISAR) is an important feature for target recognition. Owing to the addition of time dimension, ISAR image time-series are effective carriers about motion features; however, motion feature extraction is not robust enough for the restriction of ISAR imaging and traditional feature point extraction methods. In this study, the authors propose a novel method called as sparse and low-rank approach for motion feature extraction from ISAR image time-series. This method first models the ISAR image time-series as a mixed matrix composed of a low-rank matrix and a sparse matrix, which correspond to the low-rank structure of the static feature points and the sparse structure of the dynamic feature points, respectively. Then these feature points can be separated by sparse and low-rank matrices decomposition to prepare for estimation of motion parameters, such as precession angle and precession period of the targets. The results of experimental validation suggest that the new approach is more robust than traditional methods.