Video Synthetic Aperture Radar enables high-resolution, continuous imaging of observed scenes under all-weather and day-night conditions. Nevertheless, video SAR image reconstruction remains challenged by substantial data volumes and high computational complexity. This study addresses these limitations by exploiting temporal redundancy in sequential frame data. Through systematic analysis of video SAR data characteristics, we formulate video SAR imaging as a sparse tensor recovery problem by introducing a tailored correlation function to leverage inter-frame dependencies. An iterative solution is derived by integrating the alternating direction method of multipliers (ADMM) and proximal-alternating inexact minimization (P-AIM) frameworks. Based on this formulation, we propose an imaging network (ViSAR-UTNet) by unfolding the iterative process into a Transformer architecture. ViSAR-UTNet comprises two core modules: a weighted self-attention (WSA) mechanism that learns inter-frame correlations and a linearized ADMM (LADMM) operator for sparse tensor recovery. By leveraging the unfolded Transformer structure, ViSAR-UTNet effectively exploits data redundancy, thereby enabling high-quality video reconstruction from reduced measurements. Experiments on synthetic and real datasets are conducted to validate ViSAR-UTNet. The results demonstrate enhanced reconstruction accuracy and computational efficiency of the proposed method.

Gait is an appealing biometric pattern that aims to identify individuals based on the way they walk. Gait recognition, a passive human identification technology utilized from a distance without subject cooperation, plays a considerable role in life monitoring, crime prevention, security guarantee, and other identity recognition applications. Although vision-based methods dominate the state-of-the-art field, their performance degrades under poor illumination. In contrast, radar signals are not affected by light and are more sensitive to micro-motion information. In this article, we design a Kinetic feature-based Radar-Vision fused (KiRV) gait recognition method, which leverages millimeter-wave radar echo signals and a video for illumination robust human identification. In the KiRV, we propose a novel kinetic gait feature representation framework based on radar micro-Doppler and visual optical flow information, which are the direct expressions of the gait motion process. The physical meaning of the kinetic features under the two modalities is similar, while the semantic information is complementary. Therefore, the two features can be effectively fused. To learn robust gait information, we propose two 2-D residual CNN-based lightweight backbone networks to encode the kinetic features, respectively, and further propose a two-stream cross-correlated fusion method, including radar-vision cross-correlated fusion (RVCF) and radar-vision gate unit (RVGU) modules. The RVCF adaptively adjusts the attention to radar and vision for better recognition performance, while the RVGU controls the contribution of each modality to the fused feature to improve the robustness of the model. Finally, the gait retrieval task can be achieved through the above innovative model and joint loss calculation at different feature levels. Extensive experiments are conducted in the real world and semi-simulation, demonstrating that the KiRV outperforms state-of-the-art gait recognition methods with well-illumination robustness.

Synthetic aperture radar (SAR) is commonly used for ship imaging on the sea. By using small target detection algorithms, ship targets can be highlighted under different SAR backgrounds for detection and observation. Local contrast measurement (LCM) has poor detection performance in situations with strong background noise or uneven distribution, and its results cannot preserve the shape features of the original targets well. Inspired by LCM, this paper proposes a discrete cosine transform (DCT) based local contrast enhancement detection algorithm. This algorithm enhances the target area according to AC coefficient, and experimental verification and analysis show that the proposed algorithm has better detection performance and higher level of precision than LCM in the presence of complex background noise.

In the field of polarimetric synthetic aperture radar (PolSAR) automatic target classification (ATR), convolutional neural network (CNN) based methods have excelled owing to their adept feature extraction capabilities. However, these methods heavily rely on a sufficiently labeled training dataset for superior classification performance. Limited PolSAR training samples and inevitable noisy labels often render CNNs susceptible to overfitting. To tackle this challenge, a novel PolSAR image classification method employing beta mixture model and boundary amplification is proposed. Initially, the beta mixture model is utilized to fit the loss value distributions of noisy and clean samples, enabling the exploitation of distinct characteristics between these samples for probability estimation. Subsequently, to emphasize boundary samples, the boundary is delineated and expanded using the Sobel operator, amplifying losses for samples within this expanded region. Finally, a robust classification loss function is integrated into the training process to rectify losses incurred by network predictions. Experimental validation conducted on the Flevoland dataset demonstrates that the proposed method attains state-of-the-art performance.

Automatic target recognition (ATR) based on synthetic aperture radar (SAR) images has already obtained remarkable achievements on closed-set task. However, the recognition in a real-world scenario should not only identify the known classes but also appropriately deal with the unknown ones. To this end, we propose a SAR open-set recognition method aided by hierarchically reconstructive latent representation learning. First, a unsupervised representation learning via hierarchically-fused reconstruction network (HFRNet) is proposed to complement the lost information in supervised representation and obtain a preliminary closed-set result. Then, we adopt Openmax to correct closed-set recognition scores and give the probability of being the unknown ones, realizing the effective open-set recognition on SAR images. Finally, experimental results based on the measured dataset have shown the superior performance of the proposed method.

Ship recognition in synthetic aperture radar (SAR) images is a significant and fundamental step in the maritime surveillance. However, recognition of ships inevitably faces background interference in the maritime environment. The interference guides the network focusing on useless even harmful regions. To deal with issue, a mixed attention mechanism consists of coordinate and Squeeze-and-Excitation(SE) attentions is introduced. The mixed attention can guide the network to focus more on the target region, decreasing the influence of useless interference regions. Experimental and visualize results on benchmark dataset OpenSARShip validate the effectiveness of our idea.

Target recognition is one of the most significant tasks in synthetic aperture radar (SAR) image interpretation. However, due to the varying difficulty in acquiring SAR images for different categories, SAR target recognition often encounters the issue of categories imbalance. This make majority categories contribute more to the loss than minority categories, yielding a decline in classification performance. To this end, a novel SAR target recognition approach under imbalanced categories is proposed. Firstly, focal loss (FL) is introduced to balance contributions of minority and majority categories to model optimization. Then, a first-order flatness constrained FL is devised to minimize the high generalization error effectively. Finally, a gradient norm aware minimization (GAM) algorithm is implemented to integrate first-order flatness into optimization process, yielding favorable recognition results for both minority and majority categories. Experimental results based on the moving and stationary target acquisition and recognition (MSTAR) dataset demonstrate the effectiveness of our proposed method.

Synthetic aperture radar (SAR) has been widely applied in maritime target detection. However, most existing SAR ship detection algorithms based on convolutional neural network (CNN) only use single polarization SAR images for detection, neglecting to further improve the detection performance by utilizing the rich polarization information of the SAR images. To deal with this issue, this paper proposes a Cascaded Feature Fusion Pyramid Network (CFFPN) for ship detection in dual-polarization SAR images. The CFFPN builds a cascaded feature fusion module (CFFM) to fuse the enriched polarization information in SAR images. Extensive evaluations conducted on the the dual-polarization SAR ship detection dataset showcase the remarkable effectiveness of CFFPN, achieving an average precision (AP) of 93.4%. This outperforms the other five competitive methods. Notably, CFFPN exhibits a notable improvement of 1.3% in AP compared to the second-best method.

Automatic target recognition (ATR) holds a crucial position in synthetic aperture radar (SAR) image interpretation. Despite deep learning advancements have significantly propelled SAR ATR, addressing the challenge of target recognition with a few training data remains a vital concern in SAR applications. Two main issues still exist: 1) In few-shot SAR ATR, the depth and width of CNN-based models are limited, which restricts its modeling capacity, and thus extracting discriminative generalized features remains challenging. 2) With only a few labeled SAR images, the resultant class distribution is biased due to the intra-class diversity and inter-class similarity of SAR samples, which degrades the recognition performance. To address these challenges, in this letter, we propose a novel dynamically weighted prototypical learning (DWPL) method. Firstly, to extract discriminative generalized features from SAR images, we propose a new convolutional transformer network with great capacity to capture long-range dependencies of local features, together with an effective random task augmentation strategy. Secondly, in consideration of intra-class diversity and inter-class similarity, a dynamically weighted prototypical module (DWPM) is designed to adaptively assign weights to the few labeled samples that have varying discriminative information. This enables the model to effectively explore the hidden features in few samples. Through experiments conducted on the moving and stationary target acquisition and recognition (MSTAR) dataset, our method achieves recognition accuracies of 97.22% and 92.01% for 3-way 5-shot and 3-way 1-shot SAR ATR tasks in SOC, revealing significant and robust recognition performance.

Due to its all-day and all-weather capability, synthetic aperture radar (SAR) plays an important role in many remote sensing and monitoring applications. However, conventional SAR image reconstruction methods generally perform undifferentiated imaging, complicating target detection. To address this challenge, we propose a SAR image reconstruction method based on self-attention deep prior learning for differentiated image reconstruction and target detection. The proposed method can separate the target from the clutter using their feature priors during image reconstruction, thus helping improve target detection performance. Specifically, a deep prior learning operation based on a self-attention convolutional neural network (SACNN) is proposed. SACNN can help enhance target and suppress clutter by learning both the local and global features. Finally, the proposed method is implemented through an unrolled deep network with a loss function designed to make the reconstructed image beneficial for target detection. Simulation experiments have been conducted to verify the efficacy of the proposed method.

Frequency modulation continuous wave (FMCW) radar has been paid much attention in forward-looking navigation applications because of its no-blind-range capability. However, after dechirp processing, pulse interference signals may appear in the range time domain, which seriously pollutes the whole radiation direction. In this article, a cross-domain low-rank and sparse (CD-LRS) optimization framework is proposed to enhance the angular resolution and suppress the pulse interference signals based on the scanning mode of its antenna. On the one hand, to cut off and recover the polluted signals, a low-rank spectra reconstruction approach is proposed by utilizing the low-rank characteristic of the Hankel matrix formed by the interference-rejected data in the range time domain. On the other hand, to suppress the residual interference signal and enhance the angular resolution simultaneously, an adaptive sparse reconstruction method is formed in the azimuthal time domain by adopting an alternating direction method of multipliers (ADMMs)-based solver. Compared with the traditional anti-interference methods, the proposed framework can enhance the angular resolution and suppress the interference signals based on the signal features in different domains. Simulations and experimental results are applied to verify the effectiveness of the proposed framework.

Deep learning contributes to significant improvements in synthetic aperture radar (SAR) target recognition performance. Most SAR target recognition methods are based on supervised learning and require labeled SAR data. There only exists limited labeled data due to the time-consuming and laborious work of labeling, and there is still a large amount of available unlabeled radar data. Therefore, we aim to explore whether unlabeled data can provide the network with sufficient feature information and enable the network to cluster similar target features and distinguish different target features, thereby improving the SAR target recognition performance. In this paper, we propose a new framework to train a deep neural network for SAR target recognition to eliminate the need for a large amount of labeled training data. Our idea is based on transferring knowledge from a self-supervised model, where the data can train without label information. Experiments are performed on the moving and stationary target acquisition and recognition (MSTAR) benchmark dataset, and the experimental results demonstrate the improvements in recognition performance achieved by our proposed method with limited labeled data.

This paper constructs a detection network that combines convolutional neural network and Transformer in parallel to address the challenges of multi-scale target detection in remote sensing images. The network utilizes global and local information interaction to further improve the effectiveness of multi-scale object detection tasks. Additionally, the network introduces both top-down and bottom-up pathways to fuse multi-scale information, and employs coordinate attention mechanism to perform feature selection. The proposed network is compared with some existing networks on the LEVIR remote sensing image dataset, and the results show that the proposed network achieves higher average detection accuracy in terms of multi-scale target detection, particularly for small objects.

This paper considers waveform design for MIMO radar to synthesize a desired beampattern under similarity and constant modulus constraints. Generally, the constructed framework is a complex nonconvex optimization problem, which is difficult to solve directly. To tackle this problem, we convert it into a neural network-based learning problem. In particular, an objective function is developed to characterize the similarity constraint that makes the design waveform have good characteristics similar to the reference waveform. Then, we design a joint loss function for optimizing the transmit beampattern and waveform similarity, which allows the designed waveform to have better detection performance. Numerical simulation results show that the proposed method has better performance than the existing state-of-the-art method.

Synthetic aperture radar (SAR) target recognition networks performance has been remarkably improved, posing serious exposure risks to our high-value targets. Researches have shown that it is valid to protect our high-value targets by generating adversarial examples. However, most existing SAR adversarial examples generation approaches are based on the premise that irregularly global perturbation data can be directly added to SAR images, which is difficult to implement in practice. To this end, a target partial-occlusion SAR adversarial examples generation approach is proposed in this paper. First, the target region in SAR image is extracted using the combination of OTSU algorithm and morphology operations. Then, the random search (RS) algorithm is introduced to optimize the occlusion position in the extracted target region with the constraint of occlusion area and value, so as to misclassify the SAR target recognition networks. Experimental results based on the moving and stationary target acquisition and recognition (MSTAR) dataset have shown the effectiveness of the proposed method.

Marine surveillance radar plays an important role in marine environment monitoring, however, its detection performance is often affected by sea clutter. In this paper, we consider the sea clutter suppression process as the mapping from clutter radar data domain to clutter-free radar data domain, and propose a new sea clutter suppression method based on clutter cancellation generative adversarial network (CCGAN). The proposed CCGAN contains sea clutter suppression generator (SCSG) and clutter-free domain discriminator (CFDD). With the proposed network, the clutter suppression result can be obtained. To ensure the target imformation is not affected while sea clutter is suppressed, the proposed method introduces target consistency loss in addition to adversarial loss during the training process. Experimental results have shown the proposed method can achieve excellent clutter suppression performance.

Reasonable analysis of radar interference effect is of great significance for adjusting jamming strategy in radar counter-measures (RCM). The modern battlefield is confronted with non-cooperative targets, so the conventional offline evaluation methods are difficult to apply. In this paper, a comprehensive evaluation method for radar interference effect based on the integrated cloud model is proposed. Firstly, a multi-layer index system for interference effect evaluation is established. Subsequently, the entropy method is employed to determine the weight of each indicator. To avoid the occurrence of hypertrophy as an imaginary number, the cloud parameters for each indicator are calculated using a modified inverse cloud generator. Eventually, a comprehensive assessment of the interference effect can be obtained by drawing the integrated cloud. The experimental results show that the proposed method is effective and can be applied to the evaluation of interference effectiveness in non-cooperative environments.

Automatic target recognition (ATR) is the key of synthetic aperture radar (SAR) image interpretation. Due to the superior feature extraction and target classification capabilities, deep learning has been widely used in SAR ATR fields. Most of state-of-the-art SAR ATR methods are proposed for single-view input, however, multi-view SAR images include more abundant classification features. In order to improve the SAR ATR performance, it is necessary to carry out an effective method to extract and discriminate useful features from multi-view SAR images. In this paper, we propose a new SAR ATR method based on multi-view feature extraction and discrimination network, which includes two main components: multi-view feature extraction and multi-view feature discrimination. Multi-view features can be effectively extracted from input SAR images with the feature extraction component. After that, the extracted features are fed into the multi-view feature discrimination component, which aims to gather the features of the same class and separate the features of different classes. Therefore, the proposed method can achieve good multi-view SAR target recognition performance. Experiments conducted on the moving and stationary target acquisition and recognition (MSTAR) dataset demonstrate the effectiveness of our method.

In this paper, the constrained waveform design of multiple-input multiple-output (MIMO) radar is considered to achieve transmit beampattern assignment. Firstly, we construct a framework that minimizes the spatial integrated sidelobe level ratio (ISLR) as the objective function and constrains the transmit waveforms in terms of amplitude fluctuations and similarity. To solve the resulting non-convex problem, an iterative optimization method based on coordinate descent (CD) is developed by transforming the multivariate problem into multiple univariate problems. Finally, numerical simulation results demonstrate the effectiveness of the proposed method in beampattern assignment and waveform similarity.

Multiview high-resolution range profiles (HRRPs) of aerial targets contain more target information than single-view one and will benefit accurate classification. However, feature information in HRRPs dynamically varies across different views, thus a dynamic classification framework is required to adjust the structure of the network along with the feature information variations and effectively make full use of those multiview features. To this end, we propose a dynamic-structure neural network (DSNN) with skip extraction and adaptive fusion blocks to adjust the network structure and adaptively fuse multiview features, enabling accurate aerial target HRRPs classification. In the skip extraction block, the skip gate automatically changes the block depth of each view to fit feature information variations, which ensures multiview HRRP features are dynamically exploited and extracted by the network. Then, in the adaptive fusion block, features from different views are weighted by the adaptive weight gate and effectively fused using associated attention, which further contributes to the classification. Besides, since the skip gate dynamically downsizes the extraction block for some views, the computational cost of DSNN is also reduced to some extent. Experimental results demonstrate that the proposed method has superior aerial target multiview HRRPs classification performance and computational efficiency over other state-of-the-art methods.

Maritime surveillance is indispensable for civilian fields, including national maritime safeguarding, channel monitoring, and so on, in which synthetic aperture radar (SAR) ship target recognition is a crucial research field. The core problem to realizing accurate SAR ship target recognition is the large inner-class variance and inter-class overlap of SAR ship features, which limits the recognition performance. Most existing methods plainly extract multi-scale features of the network and utilize equally each feature scale in the classification stage. However, the shallow multi-scale features are not discriminative enough, and each scale feature is not equally effective for recognition. These factors lead to the limitation of recognition performance. Therefore, we proposed a SAR ship recognition method via multi-scale feature attention and adaptive-weighted classifier to enhance features in each scale, and adaptively choose the effective feature scale for accurate recognition. We first construct an in-network feature pyramid to extract multi-scale features from SAR ship images. Then, the multi-scale feature attention can extract and enhance the principal components from the multi-scale features with more inner-class compactness and inter-class separability. Finally, the adaptive weighted classifier chooses the effective feature scales in the feature pyramid to achieve the final precise recognition. Through experiments and comparisons under OpenSARship data set, the proposed method is validated to achieve state-of-the-art performance for SAR ship recognition.

Synthetic aperture radar (SAR) plays an important role in remote sensing by providing electromagnetic images of the observation scene. The prior knowledge-based SAR image reconstruction method can reduce the requirement of data sampling ratio and improve image quality. The existing prior knowledge-based method usually uses the magnitude information of SAR images while ignoring the phase information. However, since the echo and backscattering coefficients are complex values, the phase information of SAR images will help improve the reconstruction accuracy in the image reconstruction process. To improve the reconstruction performance, this paper proposes a SAR image reconstruction and autofocus method using complex-valued feature prior. In the proposed method, the complex-valued feature prior is learned from data by a complex-valued feature projection operator (CFPO), which can characterize and extract scene features in Range-Doppler domain and 2D frequency domain. The proposed CFPO enables more efficient use of echo data and helps to improve image reconstruction and autofocus performance. In addition, the proposed method is implemented by an unfolded deep network, which enables data-driven feature learning and efficient computation. The proposed method is verified by simulated and measured data.

Despite its widespread use in Earth remote sensing, synthetic aperture radar (SAR) image reconstruction remains challenging. The difficulties mainly lie in the handling of diverse scenes and motion errors with sparsely sampled data. Existing matched filtering (MF)-based methods cannot handle sparsely sampled data, while regularization-based methods lack adaptability to scene diversity. Although deep learning-based SAR methods can deal with these two issues, their performance will be degraded by motion errors. To address this, we propose a Transformer-based SAR image reconstruction method called RATIR-Net. The proposed method can obtain SAR images of various scenes under sparse sampling and motion errors by learning the correlations between echo data. In RATIR-Net, CNN-based encoding and decoding blocks are constructed to implement azimuth processes of range profiles (RP) in the range-Doppler domain according to the MF-based method. Meanwhile, a Residual Attention Transformer (RAT) block is designed to extract correlations between RPs, compensating for information loss caused by sparse sampling and suppressing non-correlated perturbations caused by motion errors. The CNN-based encoding and decoding blocks help reduce computing costs, and the RAT block mitigates the dependence on scene features and the influence of motion errors. These make RATIR-Net efficient and effective. Simulation experiments have been conducted to verify the proposed method.

Synthetic Aperture Radar (SAR) can perform all-time and all-weather observations and is wildly used in earth remote sensing. The sparsity-driven SAR imaging methods can reconstruct sparse scenes under down-sampling conditions, but they are unsuitable for non-sparse scenes. To reconstruct non-sparse scenes from under-sampled data and further improve the utilization efficiency of sampled data, this paper proposes a feature learning and SAR imaging method and implements it through a deep network. The imaging model is constructed firstly, where a feature-based sparsity regularization term is incorporated. Then, by unfolding the iterative solution derived via the Alternating Direction Multiplier Method (ADMM) algorithm, a CNN-based deep network is proposed to solve this imaging model. In the proposed network, convolution layers are used to represent and learn the scene feature prior knowledge. Simulation experiments verify the effectiveness of the proposed method.

Synthetic aperture radar (SAR) can obtain two-dimensional images of the illuminated area, which is an important means for earth remote sensing and monitoring. However, due to the loss of azimuth data and system errors during the processing of data sampling, it is necessary to study the method for high-quality SAR image reconstruction from down-sampled data in the condition of measurement inaccuracy. Considering these factors, this paper proposes a sparsity-driven SAR imaging method based on general regularization and the sparse total least-squares (S-TLS) model and implements the method by an unfolded deep network. In the proposed method, general regularization can solve the problem of sparse sampling, and the S-TLS model is adopted to deal with measurement inaccuracy. Moreover, through the deep network implementation, the proposed is more time-efficient and can exploit more effective scene prior knowledge, making the proposed method suitable in practical applications. Experiments verify the effectiveness of the proposed method.

Remote sensing images (RSIs) play an important role in a wide range of applications. However, they are frequently contaminated by multiple kinds of noises and existing methods are mostly applied to suppressing single noise type and performs poorly for various noises. To deal with above deficiencies, we propose a learning-based multi-type noise suppressing method (MNSM). Firstly, “Parallel” denoising approach is utilized to obtain partially denoised images that supply sufficient information for the subsequent fusion task. Mean-while, the noise recognition net identifies noise type and adjusts the brightness of every partially denoised image, realizing the adaptivity for different noises. The fusion net lastly merges these images to acquire one clean image. Experimental results show that this approach obtains higher peak-signal-to-noise ratio (PSNR) than existing methods.

In the field of modern electronic reconnaissance, due to the complex electromagnetic environment and the denser pulse stream, multiple radar signals will be received simultaneously. These radar signals overlap in the time domain and frequency domain, which makes modulation recognition difficult, especially under low signal-to-noise ratio (SNR). This paper proposes a Squeeze-and-Excitation InceptionNet with an adaptive threshold (SE-IncepAtNet) to deal with the above problem. The network includes an Inception block to extract features of different receptive fields and reduces the influence of noise through a Squeeze-and-Excitation (SE) block. An adaptive threshold block is used to provide adaptive thresholds, avoiding the difficulty of threshold selection in multi-classification tasks. The simulation of five typical radar signals shows that the proposed method is robust and effective.

SAR and optical images registration is a key step for remote sensing image processing, match navigation and information fusion. Although there are many methods for SAR images registration, their performance will decrease between SAR and optical images. Moreover, these algorithms suffer from lack of matching pairs of the feature points and uneven distribution between SAR and optical images. Therefore, they cannot accurately achieve the registration between optical and SAR images. To solve the above deficiencies, we propose an efficient image registration approach based on SAR and optical image-scale invariant feature transform (SOI-SIFT). Firstly, a linear edge enhancement based on gray feature and histogram equalization is introduced. In this stage, we enhance the edge features of the image so that the number of image feature points can be greatly increased. Then, for feature points purification, we use fast sample consensus algorithm to filter duplicate and wrong matching feature points. SOI-SIFT can be more adapted to the heterogeneous image matching. Experimental results have shown the superiorities of the proposed method.

Multi-view SAR images contain richer target information than single-view, which is beneficial to synthetic aperture radar automatic target recognition (SAR ATR). It is a huge challenge to select the best observation viewpoints and the most suitable flight path for multi-view SAR ATR in an unknown environment. Therefore, we propose a multi-view SAR ATR optimal observation path planning method in this paper. The geometrical and the optimization mathematical models based on the task requirements are constructed, and the convolutional neural networks with two inputs are designed as the base classifier. An autonomous path planning method forms the best observation path planning in the absence of global information of the surroundings. Thus the selection of the optimal viewpoint for multi-view SAR ATR is solved by the path search algorithm. The multi-view SAR images are collected on the solved optimal viewpoints, and the final recognition result is obtained by the base classifiers ensemble. Experimental results based on the moving and stationary target acquisition and recognition (MSTAR) dataset have shown that the proposed method obtains superiority in optimal observation path planning.

Modern warfare has entered the era of information and networking, where electronic warfare (EW) is of vital importance. Operation mode recognition occupies an important po-sition in EW, while the overlapping waveform parameters of airborne radar operation modes make it difficult to accom-plish the recognition task in complex electromagnetic environments' especially under low signal-to-noise ratio (SNR) regions. Analyzing the time-sequential regularity of radar pulse parameters and intermediate frequency (IF) sampling signals, this paper designs a novel representation of operation modes, and proposes a multi-feature residual-and-shrinkage ConvNet (RS-ConvNet) with an attention mechanism to iden-tify multiple air-to-air modes. Simulation results show the proposed method has superior performance under low SNRs.

The ambiguity function (AF) of the transmit waveform is an important reflection for cognitive radar detection system performance, and spectral coexistence is equally critical in the current frequency-congested electromagnetic environment. In this paper, a joint optimization metric related to AF and energy spectral density (ESD) is considered to improve the probability of target detection, which is accomplished by designing the radar transmit waveform. Additionally, the unimodular constraint limited by the radar transmitter is imposed on the transmit waveform. To handle the resulting nonconvex problem, an iterative optimization procedure with a closed-form solution is developed leveraging the iterative sequential quadratic optimization (ISQO) framework. Numerical simulation results are provided to demonstrate the effectiveness of the proposed approach.

Multiple-input multiple-output (MIMO) radar beampattern design has aroused extensive attention, in view of the improved detection capability, the enhanced system adaptiveness and the controlled spatial energy distribution. In addition, the ambiguity function of transmit waveform is also an important indicator of radar system performance. Thus, jointly considering the radar beampattern design and waveform similarity constraints, this paper applies the coordinate descent (CD) algorithm framework to develop an indirect ripple control approach, where beampattern ripple suppression can effectively reduce target distortion. Numerical simulation results verify the effectiveness of the proposed method in controlling ripple and similarity.

In radar forward-looking super-resolution imaging, improving the azimuth resolution while acquiring the contour information of the target has significant research value. In this letter, an approach based on the balanced Tikhonov and total variation (TV) deconvolution is proposed for radar forward-looking super-resolution imaging. We combine the Tikhonov regularization and TV regularization to construct the objective function and resolve the respective cost function using the alternating direction method of multipliers (ADMM). In each iteration, the gradient function of the target scattering coefficient is used as the adaptive weighted parameter to control automatically the weighting between the penalty terms from TV and the Tikhonov regularization. For the target with a sharper outline, the proportion of TV regularization penalty terms is increased; for the target with a smoother outline, the proportion of penalty term from the Tikhonov regularization is enhanced. The simulation and experimental results are considered to show the effectiveness of the proposed method. Compared with traditional super-resolution imaging methods, the proposed approach has superior outline retention capacity.

Ship target segmentation plays an important role in synthetic aperture radar (SAR) image interpretation. However, existing segmentation methods for marine SAR images have the problem of inaccurate edge segmentation, a concern for real-world applications. In this letter, we propose a clustering center shifted adaptive target segmentation (CCSATS) method. Firstly, the proposed clustering center shift method is used to update the clustering centers of each iteration, which can quickly and accurately capture ship pixels. Then, based on regional homogeneity coefficients, we define a new similarity measurement criterion with two adaptive weight factors to ensure the homogeneity of segmentation results. Finally, neighborhood patches are used to represent pixel information, which can reduce the influence of speckle noise and enhance the target edge fitting ability. Our segmentation results of measured SAR images show that the proposed method effectively ensures segmentation accuracy. Compared with other existing methods, the proposed target segmentation method achieves better edge capture performance.

Synthetic aperture radar (SAR) is an important means for target surveillance through reconstructing the microwave image of the observation area. However, under the condition of low signal-to-clutter ratio (SCR), such as a strong sea clutter situation, it is difficult to surveil targets from SAR images acquired by the traditional matched filter-based imaging methods. To improve the target surveillance performance of SAR, this article proposes a target-oriented SAR imaging method, which can enhance the desired target and improve the SCR in the reconstructed SAR images. By separating the target area from the clutter area, we first establish a target-oriented SAR imaging model, where the generalized regularization is used to characterize the features of the target, contributing to the improvement of SCR in the reconstructed image. Then, the imaging model is solved through a deep network, MF-ADMM-Net, which is obtained by unfolding an alternating direction method of multipliers (ADMM)-based iterative solution. In addition, the training strategy is formulated with the consideration of complex values. Experiments are conducted to verify the performance of image reconstruction and SCR improvement of the proposed method, and comparisons show the superiority of MF-ADMM-Net in effect and efficiency.

Synthetic aperture radar (SAR) can provide high-resolution electromagnetic backscattering images of the illuminated area, playing a significant role in various applications. However, achieving focused SAR images is challenging under sparse sampling and phase error conditions. By exploiting the sparsity or compressibility priors, the state-of-the-art sparsity-driven SAR imaging methods can reconstruct images under the condition of sparse sampling. However, the handcrafted priors used in these methods limit the imaging performance, and the iterative solution schemes reduce the computational efficiency. Besides, the measurement inaccuracy introduced by the phase error also degrades the reconstruction performance of the sparsity-driven imaging methods. To address these issues, a deep network for SAR autofocus imaging is proposed, which alternately performs image reconstruction and phase error estimation. When performing image reconstruction, the sparsity-cognizant total least-square (S-TLS) model is introduced to handle the problem of measurement inaccuracy, contributing to robust reconstruction performance under the condition of phase error. During the implementation of the deep network, a feature transform operator is used to realize data-driven prior knowledge learning and overcome the limitations of handcrafted priors. Moreover, the deep network approach can significantly improve computational efficiency. Experiments on simulated and real data verify the effectiveness and efficiency of the proposed method.

Superresolution methods can be applied to real aperture radar (RAR) to improve its angular resolution by solving an inverse problem. However, traditional superresolution methods are achieved after batch data collection, which requires extensive operational complexity and storage space. To solve this problem for RAR, an online detect-before-reconstruct (DBR) framework is proposed in this article based on the sparse property of targets. First, along the range direction, each sample of the echo data is detected to reduce the computational complexity by reducing the dimension of the effective data. Second, along the azimuth direction, a data-adaptive online processing structure is proposed to reduce the storage requirement for the angular superresolution problem. Finally, within the online processing structure, a target data-adaptive updating strategy is proposed to reduce the number of iterations for each target grid. The online DBR-based framework can effectively reduce the operational complexity caused by the noise values of the echo data. Based on the proposed online processing structure, the storage requirement and the operational complexity of the angular superresolution for an RAR system can be greatly reduced without significant reconstruction performance loss. The results of simulations and experimental data verify the proposed framework.

Angular resolution of real aperture radar (RAR) can be improved using deconvolution methods to achieve enhanced target information based on the convolution relationship between target scatterings and an antenna pattern. However, depending on the wide scanning scope and dense sampling angular interval, the computational complexity of the deconvolution methods will drastically increase as the dimension of azimuthal data increases. In this paper, to efficiently improve the angular resolution of RAR, a generalized adaptive asymptotic minimum variance (GAAMV) estimator that relies on a normalized projection array (NPA) model is proposed. On the one hand, the traditional convolution model of RAR is transformed into an NPA model to compress the data dimension. The proposed NPA model can normalize the signal model to make it independent of the sampling parameters. On the other hand, based on the NPA model, a GAAMV estimator is proposed to efficiently reconstruct the targets by adaptively updating each grid. Moreover, the penalty parameter is extended as a generalized case to improve its adaptability to different scenes. Based on the proposed model and method, the computational complexity can be decreased, especially for high-dimensional azimuthal data. Simulations and experimental data verify the proposed model and method.

Waveform design has become an attractive topic in the field of colocated multiple-input multiple-output (MIMO) radar that allows antennas to transmit different waveforms. Waveform properties of MIMO radar in space, time and Doppler domains determine the performances of resource utilization, interference suppression, and moving target detection. Therefore, simultaneous optimization of multi-domain properties through waveform design is significant to improve the performance of MIMO radar. In this paper, a novel MIMO radar waveform design framework that constrains the beampattern while maximizing the similarity between the designed and desired waveforms is proposed for simultaneous space-time-Doppler domain optimization. To solve the resulting multi-constraint non-convex problem, an efficient beampattern control and similarity maximization (BCSM) algorithm is developed and its convergence is demonstrated. Especially, the coupling problem due to the similarity constraint is handled by transforming the number domain and introducing the proximal algorithm. While reducing the target distortion in mainlobe region and interference in sidelobe region, the proposed method can also maximize the similarity of MIMO transmit waveforms. Numerical simulation results, apart from verifying that the proposed method outperforms existing methods in space-time-Doppler domain, also illustrate the robustness of proposed method in terms of mainlobe width and desired peak sidelobe level (PSL).

SAR image segmentation is a key step of SAR image interpretation, boosting target detection and recognition. Since similar targets may exist in complex and changeable scenes, under-segmentation and over-segmentation often occur in SAR image segmentation. To solve the above deficiencies, we propose a superpixel aggregation method based on multi-direction gray level co-occurrence matrix (GLCM) for SAR image segmentation. Firstly, a linear similarity judgment based on gray feature and spatial distance of pixels is introduced. In this stage, we expand the search range of clustering centers and add constraints to reduce the deviation, so as to alleviate over-segmentation. Then, for the spatial adjacent su-perpixels, we use multi-direction GLCM to measure texture similarity between them, merging homogeneous superpixel-s to solve under-segmentation. Experimental results based on satellite-borne SAR images from different scenes illustrate that the proposed method performs well with excellent pixel accuracy, effectively solving under-segmentation and over-segmentation.

Synthetic Aperture Radar (SAR) image is highly useful in civilian and military fields, such as ship detection, maritime search and rescue. Considering the target detection from SAR image, we propose a SAR image reconstruction and target extraction method via low-rank and sparsity constrains from under-sampled data. Firstly, the low-rank and sparsity constrains are incorporated into the SAR image reconstruction model, and the objective function is established based on Robust Principal Component Analysis (RPCA) theory. Then, the Augment Lagrange Multiplier (ALM) algorithm is used to transform this objective function to a convex optimization problem. Lastly, SAR image reconstruction and target extraction are obtained by Alternating Direction Method of Multipliers (ADMM) algorithm. The simulations are conducted to verify the effectiveness of the proposed method.

Convolutional neural networks (CNN) show superior potential in synthetic aperture radar automatic target recognition (SAR ATR). However, due to the difficulty of obtaining SAR images and the scarcity of labeled SAR images, supervised learning has poor performance in this area and is not widely applicable. To address this problem, a semi-supervised conditional generative adversarial network with a multi-discriminator (SCGAN-MD) is proposed in this paper. In our method, a conditional generative adversarial network (CGAN) is adopted with two discriminators for training the generated images and predicting the labels for unlabeled samples. Compared with other semi-supervised learning-based methods, our proposed method has more accurate image generation capability and can achieve improved recognition accuracy of SAR ATR. Experiments on the Moving and Stationary Target Acquisition and Recognition (MSTAR) database indicate that the proposed method can effectively improve the recognition accuracy and robustness of the network with a small number of labeled samples.

In the field of radar electronic reconnaissance, radar signal recognition is a key technology. In the actual task, part of the signals to be recognized may come from new types of emitters, which can not be identified directly by the existing recognition system. In order to get the ability to recognize new categories, it is necessary to analyze the unrecognized samples for incremental learning. In this paper, a new categories identification method based on reliability test is proposed. Firstly, an existing clustering method is used to label the unrecognized samples, and then the reliability test criteria are designed, including quantity criterion, distance criterion and frequency criterion, to screen the clustered sample points. The proposed method provides better data support for incremental learning in radar signal recognition. Simulation results show the effectiveness of the proposed method.

Marine surveillance radar can monitor the marine environment in all-weather conditions, but the presence of sea clutter will seriously affect its target detection performance. In this paper, we proposes a sea clutter suppression method based on machine learning that contains two pairs of generative adversarial networks (GANs), in which one GAN is used to learn the mapping relationship of sea clutter suppression, and the other is used to ensure the performance of clutter suppression. Matching loss is proposed to preserve clutter suppression performance. Experimental results have shown the superior performance of the proposed method in improving the signal-to-clutter ratio (SCR) and the stability of clutter suppression.

Designing radar waveforms with desired autocorrelation properties is a key point in the development of cognitive radar. To solve the problem of concealing weak targets by strong targets in detection, we consider minimizing the weighted integrated sidelobe level (WISL) metric in frequency domain where the weak targets are located. In order to directly solve the complex non-convex optimization problem, an iteration algorithm based on the general framework of the iterative sequential quartic optimization (ISQO) algorithm that can guarantee fast convergence to a static point is developed. Numerical simulations are provided to assess the effectiveness of the proposed algorithm.

Radar antenna can work in scanning mode to obtain a wide region observation. However, for the targets located at the scene edge, the targets are only swept by less than half of the radar beam. Therefore, the scene edge targets are recovered distortedly using the conventional angular super-resolution methods. To keep the performance of recovered targets in the full scene, in this paper, a data extrapolation-based parallel iterative adaptive approach (PIAA) is proposed. First, we analyze the cause of scene edge target distortion. Then, the echo data is extrapolated by half of the radar beam to compensate the unobserved data. Last, a parallel iterative adaptive approach is proposed to recover the targets efficiently. Simulation data is applied to verify the proposed method.

Due to the difficulty of sea scene synthetic aperture radar (SAR) trial, SAR image simulation for sea-ship scene is vitally important for the research of sea remote sensing and surveillance. In this paper, a new SAR image simulation method for sea-ship scene is proposed. Firstly, the geometrical models of sea surface and ship target are obtained through sea spectrum and CAD modeling technology respectively. Then the SAR image intensity data of sea surface is calculated by small perturbation method (SPM) and velocity bunching (VB) theory, meanwhile the radar cross section (RCS) data of ship target is computed through physical optics (PO) method. Finally, the SAR image of sea-ship scene is generated by SAR imaging method after transforming image intensity data and RCS data to the same spectrum domain. The simulation result has verified the effectiveness of the proposed method.

Identifying the aspect for a given target is an important issue in synthetic aperture radar (SAR) image interpretation. A new SAR target aspect identification method based on machine learning theory is proposed in this paper. First, the aspect angles of the SAR target are discretized, and the spatial relationships of the neighborhoods of the SAR target samples are established. Then an optimal linear mapping is solved based on the proposed subspace aspect discriminant analysis. The samples will be projected into a low-dimensional space and be of a better aspect identifiability than in their original space. Finally, the projected samples are fed into a multilayer neural network, and the aspects of the SAR targets will be indicated. Experimental results have shown the superiority of the proposed method based on the moving and stationary target acquisition and recognition (MSTAR) data set.

Bistatic synthetic aperture radar (SAR) can provide additional observables and scattering information of the target from multiple views. In this paper, a new bistatic SAR automatic target recognition (ATR) method based on multi-input deep convolutional neural network is proposed. The geometry of the multi-view bistatic SAR ATR is modeled, and an electromagnetic simulation approach is utilized as an alternative to generate enough bistatic SAR images for network training. Then a deep convolutional neural network with multiple inputs is designed, and the features of the multi-view bistatic SAR images will be effectively learned by the proposed network. Therefore, the proposed method can achieve a superior recognition performance. Experimental results have shown the superiority of the proposed method based on the electromagnetic simulation bistatic SAR data.

Oil spill will cause severe ecological disasters and enormous marine environment damages. we consider a robust and fast oil spill candidate detection problem for oil spill recognition systems for synthetic aperture radar (SAR) imagery. In this paper, we propose a automatic detection method called thresholding-guided maximally stable extremal regions (T-GMSERs) algorithm. First, thresholding approach is utlized to learn model parameters automatically. Candidate regions are extracted by using the maximally stable extremal region (MSER) detector. Then, we label each candidate region to obtain a binary potential target pixel map. Finally, the detection results are acquired by maximal stable criteria from the corresponding region map. Simulation based on satellite-borne data illustrates that the proposed algorithm obtains more precise detection performance without increasing the computational complexity.

It is a feasible and promising way to utilize deep neural networks to learn and extract valuable features from synthetic aperture radar (SAR) images for SAR automatic target recognition (ATR). However, it is too difficult to effectively train the deep neural networks with limited raw SAR images. In this paper, we propose a new approach to do SAR ATR, in which a multiview deep learning framework was employed. Based on the multiview SAR ATR pattern, we first present a flexible mean to generate adequate multiview SAR data, which can guarantee a large amount of inputs for network training without needing many raw SAR images. Then, a unique deep convolutional neural network containing a parallel network topology with multiple inputs is adopted. The features of input SAR images from different views will be learned by the proposed network layer by layer; meanwhile, the learned features from the distinct views are fused in different layers progressively. Therefore, the proposed framework is able to achieve a superior recognition performance, and requires only a small number of raw SAR images for network training samples generation. Experimental results have shown the superiority of the proposed framework based on the Moving and Stationary Target Acquisition and Recognition data set.

This paper describes a new technique for generating bistatic sea clutter returns based on the compound K-distribution model for clutter amplitude statistics. The technique adopts the computational electromagnetic (CEM) method to calculate bistatic sea clutter reflectivity by the given bistatic geometrical relationship, aiming at obtaining the parameters of the distribution. Then the theory of spherically invariant random processes (SIRP) is used to generate the returns of the bistatic sea clutter following compound K-distribution. This study can be used to evaluate the bistatic radar signal model and predict system detection performance in the sea clutter environment. Simulation results verify the proposed technique.

Recognizing the category attributes from the real world targets is one of the most challenging and attractive fields in synthetic aperture radar (SAR) application. It is an important issue to explore the spatial distribution characteristics of multi-aspect angle imagery in synthetic aperture radar automatic target recognition (SAR ATR). In this paper, we will research the spatial structure of multi-aspect angle SAR imagery through a visualization approach with real SAR data. Based on nonlinear dimensionality reduction, the representation of SAR samples is revealed in the low-dimensional Euclidean space, and the the nonlinear manifold distribution of multi-aspect angle SAR imagery is discovered. Besides, the regularity of that spatial distribution is summarized, i.e. the intrinsic structure of SAR images is parameterized by the aspect angles. The results of our research can provide a theoretical basis for SAR image classification and recognition algorithm designing.

Target image database is of great significance in SAR automatic target recognition (ATR). Recently, some convenient and low cost approaches of database simulation were proposed. However, the similarity between virtual SAR images obtained by these simulation approaches and real SAR images is still under study. To solve this problem, we will model the virtual target with three-dimensional (3D) modeling methods, and acquire SAR image via the simulated RCS data which is generated by computational electromagnetic software. Then, we propose a method to measure the similarity between the virtual and real SAR images, which provided better support for data training and recognition of virtual target. Experiment results demonstrate the formation of the virtual SAR images and validate the effectiveness of our proposed method.