Target rotation estimation for inverse synthetic aperture radar with sparse aperture and larger rotation angle

The success of inverse synthetic aperture radar (ISAR) imaging for non-cooperative target depends on accurate estimation of relative motion parameter, especially the rotation parameters. In sparse aperture and larger rotation angle configuration, the existing methods suffer too much computation or applicable condition violation. In this paper, a neural network based method is proposed to estimate the rotation parameters and it transforms the expertise hidden in the echo data generated based on the expert knowledge to the final estimator via the training process. The experimental results that the net can provide accurate estimation for echo data with appropriate SNR and the estimated parameters can help to improve the focus of imaging algorithm. Lots of examples have illustrated that the network can recognize the essential relation between the echo and the rotation motion partially.