To address the challenge of catastrophic forgetting in synthetic aperture radar (SAR) image recognition caused by viewpoint-sensitive, high-interference samples encountered in dynamic environments, we propose a lightweight and efficient online class-incremental learning (OCI) framework named Interference-Aware Replay with Dynamic Review for SAR Target Recognition (IAR-DR). Based on the experience replay (ER) mechanism, a Maximally Interfered Retrieval (MIR) strategy is designed to prioritize the replay of high-interference samples by measuring loss changes before and after model updates, thereby preserving decision boundaries under viewpoint variation. A Review Trick (RT) mechanism is further introduced to periodically revisit all buffered samples with a low learning rate, which complements MIR by reinforcing global feature retention and enhancing long-term memory stability. The combination of MIR and RT achieves a synergistic balance between local discrimination and global generalization, mitigating the forgetting effect while maintaining the efficiency. Extensive experiments conducted on the MSTAR and Bistatic MiniSAR datasets demonstrate that the proposed IAR-DR framework maintains high recognition accuracy while achieving a forgetting rate as low as 6.92% in ablation studies, and improving retention by 4.7% over recent SAR class-incremental methods.

The neuromorphic sensor’s event-based data output offers significant benefits, including minimal data redundancy and exceptional time resolution, which guarantee low power consumption and heightened sensitivity during the data acquisition process. Spiking neural network (SNN), with its inherent event-driven characteristic, is well-suited for processing event-based data, and its spike-based computing mechanism enhances the efficiency of data processing. Recent studies are exploring the integration of brain-inspired hyperdimensional computing (HDC) with SNN to leverage HDC’s advantages, including the low inference and training complexity, aiming to further reduce hardware overhead associated with SNN deployment. However, existing works have not effectively harnessed the information output by SNN during hyperdimensional encoding, leading to considerable area and energy overhead. In this article, an efficient neuromorphic HDC method is proposed, featuring a simplified hyperdimensional encoding approach that considers the temporal dynamics of SNN. In addition, a lightweight accelerator design matching the proposed method is also given. Experimental results show that the proposed accelerator achieves over 50% area reduction and reduces energy consumption by 20%–90%.

The conventional direction-of-arrival (DoA) estimation approaches are effective only when the line-of-sight (LoS) link is available. In non-line-of-sight (NLoS) scenarios, it is challenging to effectively obtain the directional information of targets due to the uncontrollability of signal reflections from NLoS links. To handle this issue, a novel reconfigurable intelligent surface (RIS)-enabled gridless DoA estimation system for NLoS scenarios is proposed, where the RIS establishes a virtual LoS link between the base station and targets. First, considering the minable statistics of the signal, the RIS-enabled signal model in the covariance domain with a limited number of receiving antennas is proposed to help reduce resource consumption. Next, we estimate the noise variance by constraining the Frobenius norm of the measurement error matrix to enhance the robustness to noise. Then, we reconstruct the Hermitian Toeplitz matrix by addressing the atom norm minimization (ANM) problem on the covariance-noiseless matrix. To reduce the computation, an efficient iterative approach is designed via the alternating direction method of multipliers . Furthermore, this system’s Cramér–Rao lower bound is derived, which is further exploited as the DoA estimation’s reference bound. Numerical experiments validate the superiority of the proposed system over the benchmark in terms of computational efficiency and estimation precision.

With low complexity and robustness, hyperdimensional computing (HDC) has become a promising paradigm for edge-side applications. HDC employs hypervectors (generally with 2–10 K dimensions) to represent input samples, and performs logical operations in hyperdimensional space to complete perceptual tasks. Compared to deep neural network (DNN), HDC is more suitable for lightweight edge-side applications (i.e., speech, activity recognition), due to its low complexity and less computational scheduling. However, existing HDC’s querying process relies on trained class hypervectors, resulting in on-chip storage and transmission overhead which limits the application of ASIC-based or FPGA-based HDC accelerators in embedded systems. In this article, a logic-aggregation-based query method called LAHDC is proposed to eliminate such overhead. In addition, an ultratiny HDC accelerator design matching LAHDC is also proposed, as well as an automated tool to search for optimal structure and generate hardware design code. Experimental results show that, compared to existing ASIC-based HDC accelerators, the proposed design reduce the area/energy by more than 95%/80%.

Due to the characteristics of light weight, low cost, and high resolution, millimeter wave (MMW) multiple-input multiple-output (MIMO) radars are widely applied in remote sensing and automotive systems. The MMW MIMO radar orthogonal waveform design is a key issue based on dechirp-on-receive technique to acquire high degree of freedom (DOF). In this paper, we propose an optimized interleaved orthogonal frequency division multiplexing (I-OFDM) chirp waveform design scheme using unequal sub-chirp duration and sparse sub-band constraint to further reduce the mutual interference (MI) between waveforms, and analyze the orthogonality of the original and optimized I-OFDM chirp waveform for MMW MIMO radar based on dechirp processing from various aspects. The simulation results show the effectiveness of the proposed method.

This paper studies super-resolution harmonic retrieval for strictly non-circular coherent signals. We develop gridless sparse representations of both their covariance and pseudo-covariance matrices over a common matrix-form atom set. This enables the decoupled atomic norm minimization (D-ANM) technique to exploit the sparsity of the covariance and pseudo-covariance matrices jointly. Further, by effectively utilizing the inherent mutual coupling characteristics between the covariance and pseudo-covariance matrices, additional constraints are properly imposed to reflect and enforce desired structure information represented by such matrices and their augmented matrix. It leads to a novel structure-based sparse optimization method, called double decoupled atomic norm minimization (DD-ANM). In addition, performance analysis is provided for the proposed DD-ANM method in practical settings. Simulation results reveal that the proposed DD-ANM outperforms the benchmark methods in terms of lower estimation errors.

Due to the target’s radar cross Section (RCS) changes with viewing angle and it is difficult to obtain target’s synthetic aperture radar (SAR) images of all the viewing angles, the training samples of SAR target recognition are always incomplete in the dimension of viewing angle. Aiming to solve the problem of views lacking in SAR image training samples, this letter proposes a multiview feature center (MVFC) sample expansion method. It is based on finding the angle-sensitive combination feature center of adjacent views. Through extracting combination features, it changes the image samples into feature dimension. Based on the correlation analysis, each sample’s similar samples could be found, and their equivalent centers are used as new sample features. By adding these centers, the sample amount could be doubled in feature dimension. At last, classifier could be trained by using the novel training sample to get better performance. This method transformed the image sample expansion problem into the feature sample expansion and used the multiview equivalent center to double the effective training sample. Experiments based on the moving and stationary target acquisition and recognition (MSTAR) dataset showed that the proposed method has higher recognition accuracy and robustness.

Brain-inspired hyperdimensional computing (HDC) is a new computational paradigm that encodes input sample into a hypervector (generally with dimensions of$2K-10K$), and performs simple arithmetic and logic operations in the hyperdimensional space to complete perceptual tasks like human brain. Due to its simplicity, interpretability, and robustness, HDC has gradually become a competitor and substitute for deep neural network (DNN) in many tasks. However, there exists an accuracy gap between existing heuristic HDC algorithms and DNN in computer vision tasks, as existing encoding methods have difficulty in filtering out large amount of background and noise in the images, and effectively extracting the spatial structure features of images. In addition, the existing hardware for HDC deployment mainly focuses on in-memory computing (IMC), application specific integrated circuit (ASIC), or high-capacity field programmable gate array (high-capacity FPGA), which cannot meet the flexibility, small area, and low power requirements of edge-side applications. In this paper, a fully learnable HDC framework with learnable preprocessing, encoding and querying, is proposed to boost the accuracy in computer vision tasks, as well as an ultra-tiny accelerator based on edge-side FPGA which matches the proposed framework. Experiments show that on multiple commonly-used image datasets, the proposed HDC framework has an average computation reduction of 80% compared to other most advanced strategies, while achieves a 1.2% accuracy increase. Evaluation on edge-side FPGA shows that compared to other FPGA based state-of-the-art designs, the proposed accelerator saves more than$10\boldsymbol{\times}$hardware resource and power consumption.

With the development of miniaturized millimeter wave (MMW) frequency-modulated continuous-wave (FMCW) radar, the dechirp-on-receive technique has been widely used. Due to the limitations of highly integrated radar hardware, it is difficult to further increase the bandwidth of the transmitted signal. Therefore, enhanced range resolution in MMW synthetic aperture radar (SAR) imaging can be achieved only thanks to suitable post-processing. In this paper, we propose a method for range resolution enhancement based on the principle of wavenumber shift with application to cross-track miniature MMW multiple-input and multiple-output (MIMO)-SAR systems. An improved orthogonal waveform design scheme of multi-subband chirp waveforms with chirp rate changes between waveforms is proposed, which is suitable for dechirp processing. In addition, given the constraint of the position of the equivalent SAR platform of MIMO-SAR, which leads to the lack of range-dimensional spectrum, a spectral data interpolation method based on autoregressive (AR) modeling in the time-frequency (TF) domain is proposed. The effectiveness of the proposed method is verified by numerical simulation and experimental data processing.

This paper proposes a super-resolution harmonic retrieval method for uncorrelated strictly non-circular signals, whose covariance and pseudo-covariance present Toeplitz and Hankel structures, respectively. Accordingly, the augmented covariance matrix constructed by the covariance and pseudo-covariance matrices is not only low rank but also jointly Toeplitz-Hankel structured. To efficiently exploit such a desired structure for high estimation accuracy, we develop a low-rank Toeplitz-Hankel covariance reconstruction (LRTHCR) solution employed over the augmented covariance matrix. Further, we design a fitting error constraint to flexibly implement the LRTHCR algorithm without knowing the noise statistics. In addition, performance analysis is provided for the proposed LRTHCR in practical settings. Simulation results reveal that the LRTHCR outperforms the benchmark methods in terms of lower estimation errors.

Although numerous classification algorithms based on deep learning can effectively extract valuable features from synthetic aperture radar (SAR) images to improve the accuracy of SAR automatic target recognition, most of them have rigorous constraints in reality and fail to obtain satisfying results with limited training samples. To address the above issues, we propose a novel multi-view attention capsule network for SAR target small sample recognition. In our method, after gaining the primary capsules of SAR images under different views, a capsule-based view attention module is designed to enhance the feature relations between different views. Then a joint dynamic routing mechanism is adopted to further capture the robust inter- and intra-image spatial relations to generate SAR capsules specialized in the final classification. Especially, to alleviate the impact of SAR target angle sensitivity on recognition, rotated cropping is applied to the original SAR images in advance. Finally, experimental results on the moving and stationary target recognition (MSTAR) and OpenSARShip dataset have demonstrated the superiority and robustness of the proposed method upon limited labeled samples.

An off-grid direction-of-arrival (DOA) estimation method based on block sparse representation is proposed to localize the strictly noncircular (NC) sources utilizing an extended transformed nested array (ETNA). This novel off-grid DOA estimation algorithm effectively promotes spatial distribution information mining. Furthermore, it is conducive to providing stable signal recovery, which refines the DOA estimation precision with interpolation over a coarse grid. We then combine the above algorithm with the designed ETNA to improve the detection performance. The ETNA is an optimal displacement on the existing TNA, which enlarges the degree of freedom (DOF) and lengthens the maximum contiguous segment from the derived virtual array. Simulation results demonstrate its superiority in estimation performance and DOF.

The orthogonal waveform diversity design and efficient hardware implementation are important issues in miniature multiple-input multiple-output (MIMO) radars. The orthogonal frequency division multiplexing (OFDM) chirp waveform has received attention recently because of its large time-bandwidth product, constant modulus, no range-Doppler coupling, good orthogonality, and good Doppler tolerance. The dechirp-on-receive technique can reduce the amount of raw sampled data in near-field miniature millimeter-wave (mmW) MIMO radar detection and synthetic aperture radar (SAR) imaging. However, most of the current waveform diversity design methods are based on general matched filtering (MF). In this paper, the possibility of using the traditional OFDM chirp waveform for dechirp processing at the receiving end of MIMO radar is analyzed. Then, the results of different configurations of chirp rates within and between transmitted waveforms for different signal processing procedures are investigated. A novel dechirp-based OFDM chirp waveform diversity design method for MIMO radar is proposed, and the results of the waveform design are given. Numerical results, such as pulse compression (PC) results, dechirp ambiguity function (DAF), SAR imaging processing, etc., and experiments verify the effectiveness of the proposed methods.

Deep learning models have achieved remarkable performance in synthetic aperture radar (SAR) target recognition. However, the accuracy of these methods is sensitive to the hyper-parameters and the traditional backpropagation is time consuming. In this letter, we proposed a modified convolutional random vector functional link (IntCRVFL) network for SAR target recognition, which can simplify the SAR target recognition system. The CRVFL network consists of a convolutional neural network and an RVFL network. First, the fixed convolutional layers with randomly initialized parameters extract SAR image features and then the RVFL network performs target recognition. Especially, inspired by hyperdimensional computing, the activations of the hidden layer are obtained through a new encoding manner. Besides, only the connections between hidden and output layers need to train by a closed-form solution for the ultimately precise target recognition. The experimental results on the moving and stationary target acquisition and recognition (MSTAR) dataset demonstrate the proposed IntCRVFL network can obtain a satisfying accuracy with a faster speed.

Deep-learning-based methods have obtained satisfying results in polarimetric synthetic aperture radar (PolSAR) image classification. However, these methods require large numbers of labeled samples, which are usually time-consuming and high-priced for PolSAR images. To address this issue, a semisupervised method based on a 3-D convolutional neural network (3-D-CNN) using pseudo labels (PL-3-D-CNN) is proposed. First, the coherency matrix of PolSAR data is converted into a 6-D real-valued vector by a unitary transformation. Then, the K-means algorithm is utilized for generating pseudo labels. After that, labeled samples and pseudo labeled samples are fed into the PL-3-D-CNN model to extract supervised and unsupervised features. Finally, the supervised and unsupervised features are combined to improve classification accuracy. The proposed method is tested on both AIRSAR and RADARSAT-2 data sets. The results show that the proposed method is an effective method for PolSAR image classification and shows good performance under a small number of labeled samples. The source code for the PL-3-D-CNN model is available athttps://github.com/fangzheng-nuaa/PL-3D-CNN.

Polarimetric synthetic aperture radar (PolSAR) image classification is an essential part of PolSAR image interpretation. In recent years, convolutional neural networks (CNNs) have made significant advances in PolSAR image classification. However, the current CNN-based methods ignore complementary information among different feature maps and correlations between elements of coherence matrix, which can provide discriminative information for classification. Besides, the phase information contained in the complex-valued (CV) coherence matrix cannot be extracted effectively. In this letter, a stacked CV convolutional long short-term memory (ConvLSTM) network called CV-ConvLSTM is proposed for PolSAR classification. Compared to existing methods, CV-ConvLSTM can extract complementary information among different feature maps and utilize the dependencies of elements in the coherency matrix, which can improve the performance of classification. In addition, the CV operations are added to the network, in which phase information is used for better classification. The experimental results of two widely used PolSAR datasets demonstrate that CV-ConvLSTM can obtain superior performance compared with existing CNN methods.

To improve the classification performance of the convolutional neural network (CNN) for polarimetric synthetic aperture radar (PolSAR) images with limited labeled samples, this letter proposes a memory CNN (MCNN) for semisupervised PolSAR image classification using both the labeled and unlabeled samples. Specifically, the MCNN introduces a memory module to realize an assimilation–accommodation interaction between the network and the module in the model training process. Compared with the traditional CNN-based methods, the advantage of the introduced interaction mechanism can exploit the memory information during the model training including both the learned feature representation and the model inference uncertainty. Under the framework of memory mechanism, the semisupervised learning can be implemented simply and effectively by introducing an unsupervised memory loss. We evaluate the proposed method on three benchmark PolSAR data sets. The experimental results show the advantages of the MCNN over the supervised, semisupervised, and unsupervised methods in the PolSAR image classification with limited labeled samples.

In recent years, convolutional neural networks (CNNs) have been successfully applied to inverse synthetic aperture radar (ISAR) sparse imaging because of their powerful ability in feature extraction. However, these CNNs only adopt single path feed-forward architectures and lack paths for directly transmitting original feature representations (OFRs) in shallow layers to reconstruction layers, which limits the complete reconstruction of target shape due to the underutilization of the OFRs that are efficient for recovering target details. Later, fully CNN (FCNN) introduces several skip connections (SKs) to establish the additional ways for directly passing the OFRs to the reconstruction layers. Nevertheless, the transmitted OFRs inevitably include the feature information of artifacts, which usually results the appearance of artifacts in final reconstructed target image. To address this issue, we introduce the gate units to FCNN, and refer to the improved FCNN as G-FCNN. Furthermore, the learnable gate units weight the OFRs transmitted by SKs and autonomously decide how many OFRs are transmitted further. To circumvent the shortage of the real data available for network training, we utilize the transfer learning strategy to guarantee a good performance of the G-FCNN. The imaging results of real data show that the G-FCNN-based imaging method is superior to the existing CNN-based imaging methods.

With the help of the auxiliary information provided by the appropriate prior fine spectral image (PFSI) in the same region, spatiotemporal super-resolution mapping (SSM) shows greater potential and better performance than the traditional super-resolution mapping (SM) models based on only monotemporal image. However, the temporal dependence of the existing SSM models usually describes the relationship between the coarse fractional images from original coarse spectral image (OCSI) and the fine fractional images from the PFSI, and the scale of temporal dependence information is not accurate and rich due to the different scales and properties of two fractional images. In addition, the existing SSM models usually do not consider point spread function (PSF) effect, resulting in affecting the accuracy of mapping result. To resolve the abovementioned issues, this letter proposes a general SSM model based on fine and coarse scales temporal dependence (FCSTD) by considering PSF effect. The experimental results demonstrate that the proposed model produces better mapping results than the traditional SM models, as well as the SSM models.

The orthogonal waveform design and good hardware realization of multiple-input–multiple-output (MIMO) radar have always been important research topics. The orthogonal frequency division multiplexing (OFDM) chirp waveform has received more attention because of its large time-bandwidth product, constant modulus, no range-Doppler coupling, good orthogonality, and good Doppler tolerance. Dechirp technique can reduce the amount of raw sampled data very well in near-field miniature lightweight MIMO radar detection and synthetic aperture radar (SAR) imaging. However, most of the current waveform analysis methods are based on matched filtering (MF). In this letter, an ambiguity function (AF) based on the dechirp signal processing approach to analyze the waveform performance is proposed, called dechirp ambiguity function (DAF). The pulse compression performance of the waveform itself and the level of mutual interference between the waveforms are described from the perspective of DAF. Numerical results validate reliability and effectiveness of the DAF.

Due to the integration and miniaturization trend of low-cost radar, how to realize fast synthetic aperture radar (SAR) image recognition on devices with strictly limited computational resources has become a problem worthy of attention. In this letter, we present a new SAR image recognition model based on the emerging brain-inspired hyperdimensional computing (HDC). Combined with the scattering mechanism of SAR image, we propose a new HDC encoding method called monogenic mapping, in which the monogenic feature vector generated by multi-scale monogenic representations of each image is directly mapped to hypervector elements in HDC after raising its order through a simple tensor product. This method can effectively replace the conventional HDC data encoding process and solve the tough problem that existing HDC models are difficult to deal with data in the 2D structure. Our lightweight model enables online, fast, and incremental learning of SAR image while well balancing high accuracy and efficiency. It also shows good stability under limited samples and noise corruption. Extensive experiments and comparisons with other algorithms on MSTAR public database demonstrate the encouraging performance of our work, which provides the possibility for HDC to deploy in more scenarios.

This paper proposes an efficient direction of departure (DOD) and direction of arrival (DOA) estimation method for multi-input multi-output (MIMO) systems. For uncorrelated scenarios, the redundancy of the covariance matrix is first exploited by establishing its concise representation through redundancy reduction, which transforms the original large-size covariance matrix into a smaller-size matrix without loss of useful angle information. Then, the resulting transformed matrix, which retains a salient structure, permits efficient two-dimensional (2D) angle estimators working on a reduced-size problem for DOD and DOA estimation. Compared with conventional subspace-based methods, the proposed method incorporating an appropriate 2D angle estimator is more computationally efficient and can achieve higher estimation accuracy for small numbers of snapshots and low signal-to-noise ratios, which are verified by simulation results.

Pansharpening technique is used to merge the original multispectral image (MS) with a high spatial resolution panchromatic image (PAN). Due to its robustness, the multiresolution analysis (MRA) is an important part of pansharpening. The scale regression model is effective for improving MRA. However, the existing MRA based on scale regression results into single-scale regression information, thus affecting the final pansharpening result. To address this problem, in this work, we propose a dual-scale regression-based MRA for pansharpening. First, we establish a scale regression-based model. Then, this model is improved using a high-pass modulation (HPM) injection scheme. Finally, the dual-scale information is added to the scale regression to construct the dual-scale regression for obtaining the final pansharpening result. We perform experiments using five datasets. The results show that the proposed method obtains a better pansharpening result as compared to various state-of-the-art MRA methods. In addition, the quantitative and qualitative analysis of the results shows that the proposed method achieves appropriate spatial and spectral resolution fusion. Therefore, it has a great potential in pansharpening technique.

Video synthetic aperture radar (VideoSAR) possesses the capability of imaging and continuously monitoring the scenario from a wide-aspect interval for enhancing the performance of information interpretation. In this paper, we propose a complete imaging scheme to achieve the unified video coordinate system in high-resolution airborne VideoSAR configuration. Comprehensive postprocessing video imaging (PPVI) framework built on range Doppler algorithm and range migration algorithm is elaborated especially in terms of complex measured data, which is divided into three parts for ensuring the stability of video background: full-aperture imaging, 2-D autofocus technique, and Doppler spectrum segmentation. Experimental results utilizing the measured airborne data have demonstrated the effectiveness of PPVI scheme for sequential VideoSAR formation.

Numerous synthetic aperture radar-automatic target recognition (SAR-ATR) methods require a large amount of training data. However, collecting SAR data is both expensive and complicated in practical applications. Recognition with the limited training data has become a vital issue in SAR-ATR. To solve this problem, we propose a recognition model combining probabilistic inference with meta-learning to transfer prior knowledge from simulated to real SAR data. First, we use various recognition tasks drawn from the simulated data to learn the global parameters of the model. Second, we draw new tasks from the real data and use the amortized inference to model a posterior distribution over task-specific parameters. Finally, we produce a predictive distribution indicating the confidence of the target classes. The experimental results demonstrate the superiority of the model in recognition tasks with a small amount of training data. We also show that introducing probabilistic inference can improve the prediction accuracy and prediction uncertainty of the model.

A fast and precise interpretation of SAR images is an important and challenging research topic. Some progress has been made in optical image interpretation through decoupling analysis method, while research on decoupling components of SAR images is still in a blank stage. To make an initial exploration on the component interpretation of SAR target images, we propose a new network based on a deep generative model and a new decoupling method. Due to the lack of real training samples that meet the required condition, we use electromagnetic simulation software FEKO to construct the training data sets. In our proposed method, we use the tag information of training samples to constrain the hidden variable layer and improve the structure and loss function of the residual variation autoencoder (Res-VAE) network. By optimizing the newly defined loss function, the network can get the decipherable component features and achieve component interpretation of SAR images. Our experiments verify the feasibility and practicability of the proposed network through the simulation data sets and MSTAR data sets. The results show that the proposed method is effective in interpreting the target components of SAR images.

Due to the influences of imaging conditions, spectral imagery can be coarse and contain a large number of mixed pixels. These mixed pixels can lead to inaccuracies in the land-cover class (LC) mapping. Super-resolution mapping (SRM) can be used to analyze such mixed pixels and obtain the LC mapping information at the subpixel level. However, traditional SRM methods mostly rely on spatial correlation based on linear distance, which ignores the influences of nonlinear imaging conditions. In addition, spectral unmixing errors affect the accuracy of utilized spectral properties. In order to overcome the influence of linear and nonlinear imaging conditions and utilize more accurate spectral properties, the SRM based on spatial-spectral correlation (SSC) is proposed in this work. Spatial correlation is obtained using the mixed spatial attraction model (MSAM) based on the linear Euclidean distance. Besides, a spectral correlation that utilizes spectral properties based on the nonlinear Kullback-Leibler distance (KLD) is proposed. Spatial and spectral correlations are combined to reduce the influences of linear and nonlinear imaging conditions, which results in an improved mapping result. The utilized spectral properties are extracted directly by spectral imagery, thus avoiding the spectral unmixing errors. Experimental results on the three spectral images show that the proposed SSC yields better mapping results than state-of-the-art methods.

A novel two‐phase method for two‐dimensional (2D) direction‐of‐arrival (DOA) estimation with L‐shaped array based on decoupled atomic norm minimization (DANM) is proposed in this paper. In the first phase, given the sample crosscorrelation matrix, the gridless DANM technique considering the noise and finite snapshots effects is employed to exploit the structure and sparse properties of the crosscorrelation matrix. The resulting DANM‐based algorithm not only enables the crosscorrelation matrix reconstruction (CCMR) but also reconstructs the covariance matrix of the L‐shaped array. Hence, sequentially, in the second phase, the conventional 2D DOA estimators for the L‐shaped array can be adopted for the angle estimation. With appropriate 2D DOA estimators, the resulting proposed algorithms can not only achieve better performance but also detect more source number, compared with conventional crosscorrelation‐based DOA estimators. Moreover, the proposed method, termed CCMR‐DANM, not only has blind characteristic that it does not require the prior information of source numbers but also is more efficient than the existing CCMR‐based counterparts. Numerical simulations demonstrate the effectiveness and outperformance of the proposed method.

This paper develops an enhanced low-rank structured covariance reconstruction (LRSCR) method based on the decoupled atomic norm minimization (D-ANM), for super-resolution two-dimensional (2D) harmonic retrieval with multiple measurement vectors. This LRSCR-D-ANM approach exploits a potential structure hidden in the covariance by transferring the basic LRSCR to an efficient D-ANM formulation, which permits a sparse representation over a matrix-form atom set with decoupled 1D frequency components. The new LRSCR-D-ANM method builds upon the existence of a generalized Vandermonde decomposition of its solution, which otherwise cannot be guaranteed by the basic LRSCR unless a very conservative condition holds. Further, a low-complexity solution of the LRSCR-D-ANM is provided for fast implementation with negligible performance loss. Simulation results verify the advantages of the proposed LRSCR-D-ANM over the basic LRSCR, in terms of the wider applicability and the lower complexity.

In order to address the problem that the direction‐of‐arrival (DOA) estimation is restricted by the number of array elements, this study comes up with a novel DOA estimation algorithm based on discrete‐Fourier‐transformation (DFT) and multiple regression, which focuses on the direct data domain from each array element rather than the traditional covariance domain of the whole array. In this algorithm, the authors regard each channel of the array signal model as an independent form of multiple regression and achieve the DOA by estimating the regression coefficient of the reconstructed source signal from each array element one by one iteratively. In addition, the DOAs estimated from the multiple array elements are averaged to improve the accuracy by exploiting the asymptotic normality. This algorithm is proposed, modelled, and developed theoretically at first. Then the capabilities that the authors' algorithm can estimate not only the very close DOAs since they distinguish the irrelevant sources by the DFT on temporal frequency but also the DOAs when the number of sources is larger than that of array elements are fully demonstrated by simulations. Furthermore, no requirement for the known number of sources brings the algorithm broader application foreground.

In this letter, a novel subpixel mapping (SPM) based on extended random walker (ERW) (SPMERW) is proposed. First, the resolution of the original coarse remote sensing image is upsampled by bicubic interpolation. Second, the class proportions of subpixel are produced by unmixing the upsampled image. Irregular objects are generated by adaptive segmentation of the first principal component of the upsampled image. Third, the class proportions of the object are derived by averaged fusion of the class proportions of subpixel belonging to each object in the segmentation image. Object spatial dependence including the spatial information among and within the objects is obtained by the ERW algorithm. Finally, a class allocation method based on units of the object is utilized to obtain the SPM result according to the object spatial dependence. Experimental results on two remote sensing data sets show that the proposed SPMERW outperforms the state-of-the-art SPM methods.

This paper aims at accurate channel estimation for millimeter-wave (mmWave) massive multiple-input multiple-output (MIMO) systems under practical limitations, including an arbitrary array geometry and a hybrid hardware structure. Taking on an angle-based approach, this work adopts a generalized array manifold separation approach via the Jacobi-Anger approximation, which transforms a non-ideal, non-uniform array manifold into a virtual array domain with a desired uniform geometric structure to facilitate super-resolution angle estimation and channel acquisition. Accordingly, structure-based optimization techniques are developed to estimate the channel parameters within a short sensing time. In particular, the difference in time-variation of path angles and path gains is capitalized to design a two-step scheme that can quickly sense fading channels. Theoretical results are provided on the fundamental limits of the proposed technique in terms of sample efficiency. Simulations testify the effectiveness of the proposed approaches.

This paper develops efficient super-resolution spatial channel covariance estimation techniques for millimeter-wave (mmWave) massive multiple-input multiple-output (MIMO) systems under the hybrid precoding constraint. Two structure-based optimization techniques including low-rank structured covariance reconstruction and dynamic atomic norm minimization are proposed to accurately estimate the channel covariance matrix. For computational efficiency, a fast iterative algorithm is developed via the alternating direction method of multipliers. The extension of this work to the higher-dimensional cases is also discussed. Simulation results verify the effectiveness of the proposed methods in hybrid mmWave massive MIMO systems.

In this paper, a new frame center extraction method based on Frame Maximum Likelihood Profile (FMLP) is proposed. Then Frame Maximum Likelihood Profile is introduced into the modeling of auto-encoder, and a new Frame Maximum Likelihood Profile-Trajectory Similarity Auto-encoder (FMLP-TSAE) is designed. On the basis of constraining the reconstructed spectra of all HRRP signals in each frame as close as possible to the spectrum of Frame Maximum Likelihood Profile, considering the characteristics of the trajectory continuity during actual acquisition of HRRP signals, FMLP-TSAE further constrains the error between reconstructed spectra of adjacent two HRRP signals to be the smallest, and the separability of extracted intrinsic deep features of different targets' HRRP signals is significantly enhanced. Experimental results show that the proposed Radar HRRP recognition method based on Stacked FMLP-TSAE has a good recognition performance whose average recognition rate of three aircraft targets reaches 99%.

This paper is concerned with continuous direction-of-arrival (DOA) estimation, and focused on developing gridless sparse methods to colored noise environment. We propose a two-stage gridless model based on noise suppression and sparse representation. Then the atomic norm minimization method is applied to recover the fourth-order cumulant matrix. We further extend the gridless SPICE method to the reconstruction of fourth-order cumulant matrix. The unknown DOAs are retrieved from the recovered matrix and the source number can be obtained as a byproduct. Numerical simulations lastly validate the computational efficiency of the proposed algorithms.

In this study, the authors address the problem of off‐grid sparsity‐inducing direction‐of‐arrival (DOA) estimation in the context of real‐valued fourth‐order cumulants (FOC) in the presence of spatially coloured noise. Firstly, a selection matrix is constructed to eliminate the redundant data of FOC and rearrange the data in the de‐redundant FOC matrix to facilitate real processing. Then a new virtual overcomplete dictionary is constructed with coupling symmetric property by linear transform with the selection matrix. Next, the FOC matrix is transformed into a real‐valued matrix via a unitary transformation which can be sparsely represented by a real‐valued virtual overcomplete dictionary. The real‐valued sparse model is vectorised for transforming to a single measurement vector (SMV) model, and the redundant data in the vector model is further removed by another selection matrix. Finally, an off‐grid sparse model based on the real‐valued SMV is established and solved by utilising the SBL strategy. The proposed method not only reduces the computational complexity but also obtains an extended‐aperture array with increased degrees of freedom which yields high resolution, and provides superiority in performance and robustness against coloured noise. The simulation results demonstrate the effectiveness of the proposed method.

Subpixel mapping based on the Hopfield neural network (HNN) is a technique to handle mixed pixels for obtaining the spatial distribution information of land cover. However, the original low-resolution remote sensing image may contain some uncertainties, such as the diversity of the land cover classes and the limitation of the resolution of the satellite sensor, the existing HNN is unable to fully utilize the prior information of the original image. In order to resolve this problem, an improved HNN (I-HNN) is proposed in this letter. In the proposed I-HNN, additional prior information of the original image is supplied by adding a new processing path to the existing HNN. To validate the effectiveness of the proposed method, two experiments are conducted on real hyperspectral images. The obtained results demonstrate that the proposed I-HNN outperforms the existing HNN. Moreover, the I-HNN does not require any auxiliary data.

Super-resolution mapping is an effective technique in mapping flood inundation for multispectral remote sensing image. However, the traditional super-resolution flood inundation mapping (SRFIM) is unable to fully utilize the spectral information from multispectral remote sensing image band. In order to resolve this problem, a novel SRFIM by supplying more spectral information (SRFIM-MSI) is proposed to improve mapping accuracy. In the proposed SRFIM-MSI, the spectral information from the multispectral band is calculated by the normalized difference water index (NDWI). A spectral term constituted by NDWI is added into the traditional SRFIM. The proposed method is evaluated by using two Landsat 8 OLI multispectral data from the study area in Cambodia. The obtained results demonstrate that the proposed SRFIM-MSI produces better results than the traditional SRFIM methods.

A variety of machine learning approaches have been applied to synthetic aperture radar (SAR) automatic target recognition. The performances of these approaches rely strongly on the quality and quantity of training data. In real-world applications, however, it is challenging to obtain sufficient data suitable for these approaches. To alleviate this problem, a novel deep generative model for SAR image generation is proposed, which is an extension of Wasserstein autoencoder. The network structure and reconstruction loss function of the model have been improved according to the characteristics of SAR images. The experimental results demonstrate that our model is superior to other classical generative models in SAR image generation. The generated images can be directly used as training samples, thereby extending the training data set and improving the recognition accuracy.

Focusing on the signal-to interference-plus-noise ratio (SINR) maximization in colocated multiple-input multiple-output (MIMO) radars, using the covariance matrix design of transmitted waveforms, we propose a kind of transmit covariance matrix (TCM)$\mathrm{R}_{pm}$with the form of symmetrical Toeplitz matrix, whose full rank characteristic firstly can sufficiently exploit the waveform diversity advantage of MIMO radar to further suppress the maximum number of interfering sources. Meanwhile, the positive semi-definition characteristic of$\sin((\pi/2)\mathrm{R}_{pm})$guarantees that these TCMs can be synthesized with binary phase shift keying (BPSK) waveforms in closed form. Furthermore, employing certain proposed TCM, higher SINR level can be yielded, and lower sidelobe levels (SLLs) can be obtained for the unwanted sidelobe interference suppression. Simulation results validate the better performance of our proposed TCMs in comparison with the phased array, omnidirectional MIMO radar and the recently proposed TCMs.

A two-stage method for off-grid coherent direction-of-arrival (DOA) estimation using atomic norm minimization based on the covariance matrix is proposed in this paper. In the first stage, by vectorizing the covariance matrix, a new off-grid model matched as a linear combination of two dimensional harmonic is presented, where the proposed denoising covariance matrix-based atomic norm minimization (DCMANM) is applied for the vectorized covariance matrix denoising. Then the simplified dual polynomial method (SDPM) is used for DOA estimation. Unlike most of existing methods, the proposed method requires knowing neither the number of signals nor the statistics of noise. Numerical simulations demonstrate the outperformance of the proposed method in both the angle resolution and DOA estimation precision compared to the state-of-the-art approaches.

In this letter, we explore the concept of group sparse representation (GSR) to exploit the intrinsic structure of synthetic aperture radar (SAR) image. Noting that dictionary design is a crucial factor in GSR performance, we propose an over-complete dictionary to better fit the SAR image despeckling problem. This over-complete dictionary consists of the prespecified dictionaries and learned dictionary. Different kinds of dictionaries emulate the image from different angles. In this way, we can simultaneously obtain better performance on speckle noise suppression and image detail preservation. The experimental results on real SAR images demonstrate that the proposed over-complete dictionary based on GSR can achieve more effective speckle reduction as well as image detail preservation.

This letter proposes a novel pulse accumulation scheme via structured measurement matrix in compressive sensing radar for better detection performance in the presence of high-level additive noise. The measurement matrix allows radar accumulate pulses with both coherent and noncoherent accumulation. The range migration can be accurately compensated via time delay compensation matrix based on frequency domain weighting. A modified algorithm is derived to recover the structured joint sparse vectors. By implementing numerical simulations, it is demonstrated that better detection performance and more accuracy can be achieved through the proposed scheme.

Due to large amount of collected data, image compression is both necessary and important for SAR system. Conventional image compression algorithms are incapable of attaining the desired compression while retaining the image fidelity required for processing at the ground station. In this paper, we present a ROI-based SAR image compression system for low bit rate. Region of interest is identified by a multiresolution constant false alarm rate (CFAR) detector and it is encoded with a higher bit rate than background. Performance was tested over the public MSTAR target chips. Results show that SNR of target area with proposed algorithm are higher than that with conventional algorithm and contextual information is preserved.

Multiresolution CFAR detection algorithm in wavelet domain using db4 wavelet is investigated in this paper, which can be incorporated into SAR image compression system. It is derived on a multiscale model of SAR imagery. Thanks to the structure of this model and detection in wavelet domain, calculation is much simpler than that of two parameter CFAR detection algorithm in image domain. Tests on simulated and collected image show that the proposed algorithm performs best. The algorithm is more suitable for the application in ROI-based image compression than two parameter CFAR detection algorithm in image domain.

The granular appearance of speckle noise in synthetic aperture radar (SAR) imagery can make it difficult to visually and automatically interpret SAR data. Speckle reduction is a prerequisite for many SAR image processing tasks. In this paper, a novel method of SAR image despeckling is presented that uses undecimated directional filter banks (UDFB) and mean shift clustering. The UDFB is obtained by manipulating the resampling matrices in the Bamberger directional filter banks (DFB), such that low computational complexity is preserved, while achieving shift invariance that could be useful in pattern recognition and image denoising applications. A nonparametric estimator of the density gradient is employed in the joint spatial-range domain of the directional bands obtained by the UDFB. Examples included at the end of the paper illustrate typical performance results obtained using this method.