Shichao Xiong

In inverse synthetic aperture radar image formation, conventional methods become ineffective under incomplete data condition, such as in sparse aperture scenario. Although sparse signal processing techniques have been introduced to improve the quality of ISAR image, off-grid problems and the structural characteristics of ISAR images still require further investigation to improve image quality. This paper proposes a sparse Bayesian learning method based on a one-dimensional off-grid coupled structure to effectively capture the structural prior information of ISAR images and correct off-grid errors simultaneously. First, an ISAR signal model with one-dimensional off-grid error compensation is established through first-order Taylor expansion. Second, within the Bayesian framework, the prior features of ISAR images are considered through a 1D coupled structural model. Finally, to achieve efficient inference, generalized approximate message passing (GAMP) is used to infer model variables. Experimental results on both simulated and real-world data demonstrate that the proposed algorithm can effectively preserve the structural features of ISAR images and correct off-grid errors in the case of incomplete data, while maintaining high computational efficiency.

For inverse synthetic aperture radar (ISAR) imaging under incomplete data conditions including sparse aperture (SA) and limited bandwidth (LB), the high-quality focused ISAR images cannot be obtained by traditional methods. Although various sparse signal processing methods have been proposed to improve ISAR imaging quality, they often neglect the intrinsic structural features of these images. Conventional pattern-coupled (PC) sparse Bayesian learning algorithms attempt to capture these structures by constructing a PC prior. However, these priors are often sub-optimal and computationally expensive. This study proposes a two-dimensional (2D) arbitrary PC prior that captures the 2D structural features to obtain high-resolution ISAR images within a Bayesian framework. Model parameters are iteratively estimated using variational Bayesian inference. Computational efficiency is improved by employing relaxed evidence lower bound for inversion-free Bayesian inference in complex number field and exploiting the Kronecker structure of the observation matrix. The noise distribution is extended to gaussian mixture model to improve the robustness in real-world with complex noises. Experiments with both simulated and measured data demonstrate that the proposed algorithm better retains the structural features of ISAR images under incomplete data, low signal-to-noise ratio and complex noises distribution conditions with high computational efficiency.

Traditional time-frequency (TF) analysis methods for micro-motion targets face significant challenges, including low resolution, susceptibility to noise interference, and issues arising from incomplete data. Although sparse signal processing offers a useful solution to the problem of incomplete data, it is often accompanied by substantial computational and incomplete TF curves. To address these challenges, this article proposes an efficient TF image recovery method. Firstly, all-frame integrated recovery (AFIR) signal model is proposed, based on which the TF recovery optimization problem is resolved using the complex-valued alternating direction method of multipliers. The TF image of all frames is recovered simultaneously, which can avoid vectorization and reduce both computational and storage costs. Further, the AFIR network (AFIR-Net) architecture is proposed. By unfolding the optimization iteration process, the hyperparameters becomes learnable, and a lightweight complex U-Net prior learning module is integrated, enabling the hyperparameters and image prior to be co-optimized during training, enhancing the quality of the TF images and keeping high computational efficiency. Experimental results on both simulated and measured data indicate that the proposed method can successfully and efficiently generate high-quality TF images in scenarios of incomplete data and noise.

Conventional time-frequency (TF) analysis methods deteriorate under the condition of data missing and low signal-to-noise ratio (SNR). Sparse signal processing method is an effective solution for recovering TF images from incomplete data, but it is burdened by significant computational and storage requirements. To address these challenges, this study proposed a TF image recovery method called multi-frame parallel recovery (MFPR) deep unfolded segmentation network. Firstly, TF recovery optimization problem solved by the proximal gradient method is constructed based on the MFPR signal model which can simultaneously recover TF images from all frames. This can eliminate the need for vectorization and thereby reduce computational and storage costs. Then, MFPR deep unfolded segmentation network (MDUS-Net) which contains two sub-networks is proposed to obtain complete TF image from incomplete data. The first sub-network is a deep unfolded network of MFPR, which achieves high-quality TF recovery from incomplete data. The second sub-network is a segmentation network, which achieves TF curves skeleton extraction via image segmentation technique. Experimental results on both simulated and measured data demonstrate that the proposed method can generate high-quality TF images even under conditions of incomplete data and low SNR.

Inverse synthetic aperture radar (ISAR) imaging of maneuvering targets presents a significant challenge. Maneuvering targets exhibit time-varying Doppler shifts along the azimuth, resulting in a defocused ISAR image. The Range Instantaneous Doppler (RID) algorithm can achieve instantaneous focused ISAR images of maneuvering targets. However, the performance of RID deteriorates under sparse aperture (SA) conditions and low signal-to-noise ratio (SNR). Although the SA problem can be addressed using compressed sensing-based RID, it is burdened by significant storage and computational demands. In addition, the ISAR image exhibits a block-sparse structure in the spatial domain. To efficiently obtain the focused ISAR image from sparse aperture (SA) data in a low SNR environment, we propose a fast non-uniform block SBL algorithm (FNUBSBL). The proposed algorithm reconstructs ISAR images frame by frame to reduce storage and computational demands. To exploit structural features, the sparse signal is modeled using a non-uniform block sparse prior. An inverse-free variational Bayesian inference is developed to achieve accurate inference with high efficiency. In light of the continuity inherent in the time-frequency image, a dimension reduction strategy has been proposed to mitigate computational complexity. Experimental results obtained from both simulated and measured data indicate that the proposed algorithm can efficiently generate ISAR images while effectively preserving the block structure of these images under conditions of SA and low SNR.

ISAR imaging for maneuvering targets (MT) in sparse aperture (SA) condition is a challenging problem. Range instantaneous Doppler (RID) is useful for ISAR imaging of MT through time-frequency analysis. However, the performance of RID deteriorates in SA, and frequency resolution is limited by the assumption of stationary signal in the time window. To tackle these issues, a complex value parametric efficient sparse Bayesian learning (CPESBL) ISAR imaging algorithm is proposed in this letter. In our algorithm, the one frame signal of MT ISAR imaging is modeled as the multicomponent Chirp signal. This model is solved by CPESBL which contains the complex value efficient SBL (CESBL) with low computational complexity and the Quasi-Newton method estimating the Chirp rate parameter. Then the focused ISAR image can be obtained efficiently. Moreover, a dimension shrinkage strategy is also proposed to further improve the computational efficiency considering the continuity of sequential ISAR images. With a low computational complexity, the proposed algorithm achieved the best image quality index both in simulated and measured data experiments.

Currently, the sparse imaging problem of synthetic aperture radar (SAR) is primarily addressed by compressed sensing (CS) theory, which introduces prior information into image restoration tasks via regularization. However, simple regularization constraints cannot provide the complex structural information of targets, and their denoising performance is unsatisfactory at low signal-to-noise ratios (SNRs) and sampling rates. In this letter, a novel sparse SAR reconstruction network is proposed based on plug-and-play (PnP) and approximated observation. First, a chirp-scaling algorithm (CSA) operator-derived approximated observation model is applied to reduce computational costs. Then, this sparse imaging problem is iteratively solved with a matched filter-based approximate message-passing (MFAMP) method. To overcome the limitations of prior models in existing sparse imaging methods, a PnP prior model is incorporated within the sparse reconstruction framework instead of using the ℓ ₁ sparse regularizer. Finally, the solution procedure is unfolded as a deep imaging network, dubbed as PnP-MFAMP-Net. Experimental results validate its robustness and superiority. Even at a sampling rate of 25%, PnP-MFAMP-Net can achieve a PSNR gain of approximately 15 dB compared to the AMP-Net.

In recent decades, the deception jamming approach based on digital radio frequency memory (DRFM) for inverse synthetic aperture radar (ISAR) has been a well-researched topic. Various types of jamming signals based on interrupted-sampling repeater jamming (ISRJ) can induce one-dimensional multiple high-resolution range profile (HRRP) false targets or two-dimensional realistic multiple ISAR false targets for deception. However, these existing methods generate false targets that are identical to the real target. The ISRJ false target also generates a main false target whose energy is much higher than other sub-false targets. Thus, it is easy to discover that the radar has suffered jamming. In order to generate better, more confusing jamming signals, this paper proposes a jamming method based on sub-Nyquist sampling jamming to induce realistic, multiple false targets on ISAR images. It improves a template multiplication modulated time-delay method to eliminate and add scatterers on the selected false target to change its ISAR shape. The frequency-shift parameters of the template jamming signal are analyzed and derived in detail for eliminating and adding scatterers at the specified location. Thus, multiple false targets with different ISAR shapes and similar energy are generated, which can create better deception effects. Meanwhile, this method can adjust the number of false targets and the location of changed-shape false targets. Furthermore, the resolution of the false target can adaptively change with the radar pulse width and the accumulated pulse number. The simulation results show that the proposed deception jamming strategy works.