Joint Motion Compensation Radar Imaging Based on Bi-XLSTM and BP Algorithm

Abstract

Advances in high-speed digital acquisition and storage technologies have enabled the sampling of radar echoes at intermediate frequencies. Precise gate control and recording techniques facilitate translational compensation based on echo coherence, enabling robust imaging even in low signal-to-noise ratio (SNR) conditions. Despite these advancements, existing approaches face inefficiencies in parameter estimation and encounter significant difficulties in azimuth focusing when increasing imaging resolution at large angles. To address these challenges, a novel joint motion compensation imaging algorithm that integrates bidirectional extended long short-term memory (Bi-XLSTM) networks with back projection (BP) is proposed. This approach initiates with the development of a comprehensive joint motion echo model for the target to reduce cumulative errors, thereby establishing a solid foundation for further joint motion compensation. Next, Bi-XLSTM is utilized for the efficient extraction of motion parameters from coherent echoes. The Adam optimization algorithm is then applied during the BP imaging process to jointly optimize motion and rotation parameters, targeting optimal image quality and resulting in highly focused images. Experiments conducted on point simulations, electromagnetic simulations, and real data confirm that this technique outperforms traditional methods.