High resolution is a key trend in the development of synthetic aperture radar (SAR), which enables the capture of fine details and accurate representation of backscattering properties. However, traditional high-resolution SAR imaging algorithms face several challenges. Firstly, these algorithms tend to focus on local information, neglecting nonlocal information between different pixel patches. Secondly, speckle is more pronounced and difficult to filter out in high-resolution SAR images. Thirdly, the process of high-resolution SAR imaging generally involves high time and computational complexity, making real-time imaging difficult to achieve. To address these issues, we propose a hyper-pixel high-resolution SAR imaging network (HPHR-SAR-Net) for rapid despeckling in high-resolution modes. Based on the concept of hyper-pixel techniques, we initially combine nonconvex and nonlocal total variation as compound regularization. This approach more effectively despeckles and manages the relationship between pixels while reducing bias effects caused by convex constraints. Subsequently, we solve the compound regularization model using the Alternating Direction Method of Multipliers (ADMM) algorithm and unfold it into a Deep Unfolded Network (DUN). The network's parameters are adaptively learned in a data-driven manner, and the learned network significantly increases imaging speed. Additionally, the proposed HPHR-SAR-Net is compatible with high-resolution imaging modes such as spotlight, staring spotlight, and sliding spotlight. In this paper, we demonstrate the superiority of HPHR-SAR-Net through experiments in both simulated and real SAR scenarios. The results indicate that HPHR-SAR-Net can rapidly perform high-resolution SAR imaging from raw echo data, producing accurate imaging results.

The multi-agent consensus equilibrium (MACE) mechanism, which generalizes the popular method PnP-ADMM and composite regularization in computational sensing, possesses the notable capacity to incorporate multiple priors aspect of both regularization and Plug-and-play (PnP) for improving image quality. In this work, a flexible SAR image reconstruction method based on MACE is proposed to integrate multiple regularization and PnP priors for various features enhancement. The partial-update approach and Mann iteration methods are implemented to increase the computational efficiency of the MACE-based SAR image reconstruction algorithm. A thorough analysis of the proposed algorithm’s convergence and computational complexity is provided. High-quality SAR images necessitate low ambiguity, high target to background ratio (TBR), and low coherent speckle. We therefore demonstratively integrate regularization and PnP priors for azimuth ambiguity suppression, sparsity inducing, multiple features enhancement and despeckling. The proposed method’s performance is evaluated through experiments on both simulated and QILU-1 satellite SAR data.

As a branch application of the synthetic aperture radar (SAR), coherent change detection (CCD) is commonly utilized to help differentiate specific changes from the unchanged scene. To exactly pick out man-made changes of interest from change maps, the introduction of artificial neural networks (ANNs) has significantly enhanced detection performance but taken up great computational resources, making it challenging to deploy on edge devices with limited resources and a strict power budget. Spiking neural networks (SNNs) not only infer results more energy-efficiently but also have the potential to drive ultra-low-power neuromorphic computers which ANNs lack. In this study, to detect coherent changes precisely and with less energy consumption, a robust SNN-based change analysis method is proposed to make it available to implement CCD tasks via the “WenTian I” brain-like computer. Firstly, we adopt a Poisson spike encoding method based on fuzzy c-means (FCM) clustering to pre-classify pixels and translate continuous change maps into spikes. Secondly, we input the spike images into a novel deep SNN, multi-threshold Spiking-UNet. By representing information in the form of sparse spike flows, our proposed change detection model can reduce the energy consumption of CCD while maintaining excellent detection capabilities. The corresponding experimental results validate the comparable detection effectiveness and superior energy efficiency of our SNN-based method used for SAR coherent change detection.

In automotive millimeter-wave (MMW) radar systems, achieving high-precision direction of arrival (DOA) estimation with a limited number of array elements is a crucial research focus. Compressive sensing (CS) techniques have been demonstrated to offer superior performance in DOA estimation compared to spectral estimation methods. However, traditional CS methods suffer from an off-grid effect, which causes their reconstruction results to deviate from the actual positions of the signal sources, thereby reducing the accuracy. Currently, as a gridless method, atomic norm minimization (ANM) has shown effectiveness in DOA estimation for uniform linear arrays (ULAs). However, the performance of ANM is suboptimal in non-uniform linear arrays (NULAs), and their computational efficiency is not satisfactory. In this paper, we propose a novel algorithm for DOA estimation in NULA, drawing inspiration from the alternating descent conditional gradient algorithm framework. First, we construct an atomic set based on the observation scene and select the atoms with the highest correlation to the residuals as potential signal sources for global estimation. Then, we construct a mapping function for the signal sources in the continuous domain and perform conditional gradient descent in the neighborhood of each signal source, addressing the bias introduced by the off-grid effect. We compared the proposed algorithm with ANM, Iterative Shrinkage Thresholding (IST), and Multiple Signal Classification (MUSIC) algorithms. Simulation experiments validate that the proposed algorithm effectively addresses the off-grid effect and is applicable to DOA estimation in coprime and random arrays. Furthermore, real data experiments confirm the effectiveness of the proposed algorithm.