Unblurring ISAR Imaging for Maneuvering Target Based on UFGAN

Inverse synthetic aperture radar (ISAR) imaging for maneuvering targets suffers from a Doppler frequency time-varying problem, leading to the ISAR images blurred in azimuth direction. Given that traditional methods have poor performance or low efficiency, and existing deep learning cannot effectively reconstruct deblurred retaining rich details textures, an unblurring method based on advanced Transformer structure is proposed. We first present pseudo-measured data generation DeepLabv3+ network Diamond-Square algorithm acquire dataset training with good generalization measured data. Next, locally-enhanced window block adopted enhance ability capture local context as well global dependencies, we construct novel Uformer-based GAN (UFGAN) restore textures results. The simulation experiments show proposed can achieve fast high-quality under condition of signal-to-noise ratio (SNR) sparse aperture.