Method for scatterer trajectory association of sequential ISAR images based on Markov chain Monte Carlo algorithm

摘要

Based on sequential inverse synthetic aperture radar (ISAR) images, the three-dimensional target structure can be reconstructed using the factorisation method. However, it requires accurate scatterer trajectory formation, which is difficult due to the occlusion and trajectory crossing. To address this problem, the authors propose a novel scatterer trajectory association method based on Markov chain Monte Carlo (MCMC) algorithm. First, they derive the ellipse movement characteristics of each scatterer trajectory under stationary rotational motion model of the observed target. Then, by computing the signal-to-noise ratio of the compressed echoes, the number and positions of the scatterers in each ISAR image can be extracted precisely and efficiently through two-dimensional estimation of signal parameters via rotational invariance techniques. Next, they present a Bayesian model and inference algorithm for the scatterer trajectory association problem. MCMC is applied to estimate the scatterer trajectory matrix. Particularly, they design new prior and likelihood evaluation criterions in MCMC by making use of the ellipse movement characteristics of each scatterer trajectory. Experimental results on simulated data validate the effectiveness of the proposed method.