A fast inverse synthetic aperture radar translational motion compensation method for low SNR scenes

Abstract Accurate translational motion compensation is crucial for inverse synthetic aperture radar (ISAR) imaging. At low signal-to-noise ratio (SNR), the performance of nonparametric methods is declining because the correlation of adjacent range profiles is severely destroyed by the noise. Fortunately, parametric methods can still work well at low SNR. However, existing parametric methods are time-consuming, which decreases ISAR imaging efficiency significantly. So, a fast parametric translational motion compensation method is studied for low SNR scenes in the paper, which is based on modified Adam (Madam) algorithm. By changing the updating strategy for weights and learning rate in Adam, Madam algorithm can significantly accelerate the convergence when estimating the motion parameters. To further improve the efficiency, a new objective function by combining image contrast and image entropy is proposed. When applied into measured Yak-42 data at low SNR, the proposed method can image about 10 times faster than particle swarm optimization (PSO) method and 2 times faster than Adam method. Therefore, the proposed method is an efficient translational motion compensation method for low SNR ISAR imaging.