Multipath Ghost Suppression of Extended Targets Based on SAR Parametric Sparse Imaging Model

‍ ‍Synthetic Aperture Radar （SAR） is an active remote sensing technology that acquires two-dimensional images of a target area by transmitting and receiving electromagnetic waves. In practical scenarios， particularly in urban areas， there are a large number of extended targets such as building liner edges and bridge liner structures. SAR receives signals that not only include the direct reflection of electromagnetic waves from the target but also interference echoes from multiple reflections between the target and the environment background （such as ground or water）， known as the “multipath effect”. This effect causes “ghosting” in SAR imaging. Most current multipath suppression methods are based on the point target model， which is challenging to apply effectively to solve the ghosting problem caused by extended targets. Therefore， this paper proposed a multipath ghost suppression of extended targets based on SAR parametric sparse imaging model， which can achieve high-quality imaging of straight linear extended targets in the presence of multipath effects caused by the sparse targets and environmental background. Firstly， a multipath observation model for extended targets was established based on the target’s scattering model under different echo paths， which can describe the scattering characteristics of point and extended targets under different echo paths. Secondly， the alternating direction method of multipliers （ADMM） was used to estimate accurately the scattering coefficient vectors of the target under different echo paths in the multipath observation model of extended targets. Finally， an image reconstruction method was proposed for different echo paths based on the estimated results， which can concentrate the energy of each echo path in the real target area and achieve high-quality imaging of straight linear extended targets. The effectiveness of the proposed method was verified through high-precision electromagnetic simulation data.