MSTAR Dataset Research Articles

Attention Heat Map-Based Black-Box Local Adversarial Attack for Synthetic Aperture Radar Target Recognition

Synthetic aperture radar (SAR) automatic target recognition (ATR) models based on deep neural networks (DNNs) are susceptible to adversarial attacks. In this study, we proposed an SAR black-box local adversarial attack algorithm named attention heat map- based black-box local adversarial attack (AH-BLAA). First, we designed an attention heat map extraction module combined with the layer-wise relevance propagation (LRP) algorithm to obtain the high concerning areas of the SAR-ATR models. Then, to gener- ate SAR adversarial attack examples, we designed a perturbation generator module, introducing the structural dissimilarity (DSSIM) metric in the loss function to limit image distortion and the dif- ferential evolution (DE) algorithm to search for optimal perturba- tions. Experimental results on the MSTAR and FUSAR-Ship datasets showed that compared with existing adversarial attack algorithms, the attack success rate of the AH-BLAA algorithm increased by 0.63% to 33.59% and 1.05% to 17.65%, respectively. Moreover, the low- est perturbation ratios reached 0.23% and 0.13%, respectively.

Exploring Reinforced Class Separability and Discriminative Representations for SAR Target Open Set Recognition

Current synthetic aperture radar (SAR) automatic target recognition (ATR) algorithms primarily operate under the closed-set assumption, implying that all target classes have been previously learned during the training phase. However, in open scenarios, they may encounter target classes absent from the training set, thereby necessitating an open set recognition (OSR) challenge for SAR-ATR. The crux of OSR lies in establishing distinct decision boundaries between known and unknown classes to mitigate confusion among different classes. To address this issue, we introduce a novel framework termed reinforced class separability for SAR target open set recognition (RCS-OSR), which focuses on optimizing prototype distribution and enhancing the discriminability of features. First, to capture discriminative features, a cross-modal causal features enhancement module (CMCFE) is proposed to strengthen the expression of causal regions. Subsequently, regularized intra-class compactness loss (RIC-Loss) and intra-class relationship aware consistency loss (IRC-Loss) are devised to optimize the embedding space. In conjunction with joint supervised training using cross-entropy loss, RCS-OSR can effectively reduce empirical classification risk and open space risk simultaneously. Moreover, a class-aware OSR classifier with adaptive thresholding is designed to leverage the differences between different classes. Consequently, our method can construct distinct decision boundaries between known and unknown classes to simultaneously classify known classes and identify unknown classes in open scenarios. Extensive experiments conducted on the MSTAR dataset demonstrate the effectiveness and superiority of our method in various OSR tasks.

Physics-guided interpretable CNN for SAR target recognition

Deep Learning (DL) model has been widely used in the field of Synthetic Aperture Radar Automatic Target Recognition (SAR-ATR) and has achieved excellent performance. However, the black-box nature of DL models has been the focus of criticism, especially in the application of SAR-ATR, which is closely associated with the national defense and security domain. To address these issues, a new interpretable recognition model Physics-Guided BagNet (PGBN) is proposed in this article. The model adopts an interpretable convolutional neural network framework and uses time–frequency analysis to extract physical scattering features in SAR images. Based on the physical scattering features, an unsupervised segmentation method is proposed to distinguish targets from the background in SAR images. On the basis of the segmentation result, a structure is designed, which constrains the model’s spatial attention to focus more on the targets themselves rather than the background, thereby making the model’s decision-making more in line with physical principles. In contrast to previous interpretable research methods, this model combines interpretable structure with physical interpretability, further reducing the model’s risk of error recognition. Experiments on the MSTAR dataset verify that the PGBN model exhibits excellent interpretability and recognition performance, and comparative experiments with heatmaps indicate that the physical feature guidance module presented in this article can constrain the model to focus more on the target itself rather than the background.

Leveraging Visual Language Model and Generative Diffusion Model for Zero-Shot SAR Target Recognition

Simulated data play an important role in SAR target recognition, particularly under zero-shot learning (ZSL) conditions caused by the lack of training samples. The traditional SAR simulation method is based on manually constructing target 3D models for electromagnetic simulation, which is costly and limited by the target’s prior knowledge base. Also, the unavoidable discrepancy between simulated SAR and measured SAR makes the traditional simulation method more limited for target recognition. This paper proposes an innovative SAR simulation method based on a visual language model and generative diffusion model by extracting target semantic information from optical remote sensing images and transforming it into a 3D model for SAR simulation to address the challenge of SAR target recognition under ZSL conditions. Additionally, to reduce the domain shift between the simulated domain and the measured domain, we propose a domain adaptation method based on dynamic weight domain loss and classification loss. The effectiveness of semantic information-based 3D models has been validated on the MSTAR dataset and the feasibility of the proposed framework has been validated on the self-built civilian vehicle dataset. The experimental results demonstrate that the first proposed SAR simulation method based on a visual language model and generative diffusion model can effectively improve target recognition performance under ZSL conditions.

Circular synthetic aperture radar sub‐aperture angle information complementation based on azimuth‐controllable generative adversarial network

AbstractA conditional generative adversarial network (CGAN) framework is proposed to address the issue of incomplete circular synthetic aperture radar (CSAR) azimuthal information due to motion errors. Specifically, the authors propose a novel CGAN architecture that can control the azimuth angle for arbitrary angle generation, capable of complementing missing CSAR sub‐aperture information. The network incorporates angular labels for various scenarios and integrates a dynamic region‐aware convolution (DRconv) module. Additionally, to counteract the common challenge of mode collapse in GAN training, a mode seeking regularisation technique is innovativrly introduced into the authors’ loss function. The efficacy of the proposed network is rigorously tested using both the MSTAR dataset and an X‐band SAR dataset. The results demonstrate that the authors’ network can generate high‐fidelity SAR images with controllable azimuths, closely resembling authentic images. Furthermore, the proposed method excels in complementing missing CSAR sub‐aperture information, effectively supplying the lost angular information due to motion errors. A new technical approach for SAR image generation is not only offered but it also has the potential to significantly expand SAR datasets. This advancement is expected to enhance the quality and utility of SAR imagery in applications such as surveillance, reconnaissance, and environmental monitoring.

Enhancing Transferability with Intra-Class Transformations and Inter-Class Nonlinear Fusion on SAR Images

Recent research has revealed that the deep neural network (DNN)-based synthetic-aperture radar (SAR) automatic target recognition (ATR) techniques are vulnerable to adversarial examples, which poses significant security risks for their deployment in real-world systems. At the same time, the adversarial examples often exhibit transferability across DNN models, whereby when they are generated on the surrogate model they can also attack other target models. As the significant property in black-box scenarios, transferability has been enhanced by various methods, among which input transformations have demonstrated excellent effectiveness. However, we find that existing transformations suffer from limited enhancement of transferability due to the unique imaging mechanism and scattering characteristics of SAR images. To overcome this issue, we propose a novel method called intra-class transformations and inter-class nonlinear fusion attack (ITINFA). It enhances transferability from two perspectives: intra-class single image transformations and inter-class multiple images fusion. The intra-class transformations module utilizes a series of diverse transformations that align with the intrinsic characteristics of SAR images to obtain a more stable gradient update direction and prevent the adversarial examples from overfitting the surrogate model. The inter-class fusion strategy incorporates the information from other categories in a nonlinear manner, effectively enhances the feature fusion effect, and guides the misclassification of adversarial examples. Extensive experiments on the MSTAR dataset and SEN1-2 dataset demonstrate that ITINFA exhibits significantly better transferability than the existing transfer-based methods, with the average transfer attack success rate increases exceeding 8% for single models and over 4% for ensemble models.

Application of adaptive fusion to attributed scattering centre-based reconstruction for SAR target recognition

ABSTRACT Synthetic aperture radar (SAR) measures backscattering characteristics of man-made targets and has the advantage of not being affected by weather and time. For target recognition in SAR images, the target reconstruction results based on attribute scattering centres are introduced for decision fusion. In comparison with original images, the reconstructed targets better relieve corruptions caused by noises, clutters, etc. A deep learning model, i.e. ResNet, is adopted as a basic classifier to classify both original and reconstructed images. According to the energy relationship between the reconstruction target and residual, the noise level of the original SAR image is defined. Then, the adaptive weights of original and reconstructed images are determined and a weighted decision fusion process is conducted to combine decisions from both images to confirm the target label. The proposed method is tested based on MSTAR dataset, and experimental results show its effectiveness.

Synthetic aperture radar automatic target recognition based on cost‐sensitive awareness generative adversarial network for imbalanced data

AbstractIn military contexts, synthetic aperture radar (SAR) automatic target recognition (ATR) models frequently encounter the challenge of imbalanced data, resulting in a noticeable degradation in the recognition performance. Therefore, the authors propose a cost‐sensitive awareness generative adversarial network (CAGAN) model, aiming to improve the robustness of ATR models under imbalanced data. Firstly, the authors introduce a convolutional neural network (DCNN) to extract features. Then, the synthetic minority over‐sampling technique (SMOTE) is applied to achieve feature‐level balancing for the minority category. Finally, a CAGAN model is designed to perform the final classification task. In this process, the GAN‐based adversarial training mechanism enriches the diversity of training samples, making the ATR model more comprehensive in understanding different categories. In addition, the cost matrix increases the penalty for misclassification results and further improves the classification accuracy. Simultaneously, the cost‐sensitive awareness can accurately adjust the cost matrix through training data, thus reducing dependence on expert knowledge and improving the generalisation performance of the ATR model. This model is an end‐to‐end ATR scheme, which has been experimentally validated on the MSTAR and OpenSARship datasets. Compared to other methods, the proposed method exhibits strong robustness in dealing with various imbalanced scenarios and significant generalisation capability across different datasets.

Vehicle Target Recognition in SAR Images with Complex Scenes Based on Mixed Attention Mechanism

With the development of deep learning in the field of computer vision, convolutional neural network models and attention mechanisms have been widely applied in SAR image target recognition. The improvement of convolutional neural network attention in existing SAR image target recognition focuses on spatial and channel information but lacks research on the relationship and recognition mechanism between spatial and channel information. In response to this issue, this article proposes a hybrid attention module and introduces a Mixed Attention (MA) mechanism module in the MobileNetV2 network. The proposed MA mechanism fully considers the comprehensive calculation of spatial attention (SPA), channel attention (CHA), and coordinated attention (CA). It can input feature maps for comprehensive weighting to enhance the features of the regions of interest, in order to improve the recognition rate of vehicle targets in SAR images.The superiority of our algorithm was verified through experiments on the MSTAR dataset.

A Novel Lightweight CNN for Constrained IoT Devices: Achieving High Accuracy With Parameter Efficiency on the MSTAR Dataset

A Novel Lightweight CNN for Constrained IoT Devices: Achieving High Accuracy With Parameter Efficiency on the MSTAR Dataset

A Feasibility Analysis of the Use of ISAR Training Data in Machine Learning-Based SAR ATR

Processing of synthetic aperture radar (SAR) images for automatic target recognition (ATR) is a critical application especially in military surveillance. In particular, numerous machine learning-based SAR ATR methods have been proposed for this task. However, data training and testing stages of all these methods are based on the exploitation of SAR signatures of the target under investigation. Considering the high variability of radar targets, obtaining such signature data is obviously a costly and time consuming process. In this study, therefore, a feasibility analysis of the use of inverse-SAR (ISAR) training data in SAR ATR has been made for the first time. The turntable ISAR and circular SAR images of three different vehicles are used in training and testing is performed by means of SAR images of three similar targets within the publicly available MSTAR dataset. Also, three most prominent machine learning methods, namely KNN, SVM and ANN are used in conjunction with three different feature extraction algorithms namely, GLRLM, GLSZM and GLCM. The obtained results reveal that the GLCM+SVM algorithm pair is the most effective model with 85% accuracy.

SAR-ATR with knowledge hierarchy division and information dissemination networks

Synthetic aperture radar (SAR) images possess diverse domain characteristics that describe targets from multiple perspectives, including scattering center, intensity, phase, azimuth angle, and shadow. The imaging mechanism of SAR images significantly differs from optical images due to its use of radio waves instead of visible light for imaging purposes. However, current deep learning (DL)-based automatic target recognition (ATR) approaches for SAR primarily focus on the intensity information while neglecting the comprehensive consideration of SAR domain characteristics. This limitation results in insufficient expression of target features and hampers further improvement in SAR target recognition performance. To address these issues, this paper proposes a novel SAR target recognition method called KDNets based on information dissemination networks (IDNets) and knowledge hierarchy division (KnHD). KDNets effectively enhances the feature representation ability of SAR targets by exploiting multiple domain characteristics at sample, feature, and decision levels. Specifically, IDNets incorporates scattering center and azimuth angle at sample and feature levels to extract rich multi-scale semantic features for representing SAR targets. At the decision stage, KnHD utilizes azimuth information to rapidly search and fuse prior knowledge about targets from a knowledge base to achieve interpretable and high-precision SAR target recognition. Experimental results on the MSTAR dataset and MA-SAR dataset demonstrate that our proposed KDNets can achieve state-of-the-art performance in SAR target recognition tasks, with the accuracy of 99.75 % in the MSTAR dataset and 98.61 % in the MA-SAR dataset, validating its effectiveness and superiority. Furthermore, this study highlights the significant potential and application prospects of leveraging domain characteristics in SAR-ATR tasks.

Crucial feature capture and discrimination for limited training data SAR ATR

Deep learning-based methods have demonstrated exceptional performance in the field of synthetic aperture radar automatic target recognition (SAR ATR). However, obtaining a sufficient number of labeled SAR images remains a significant challenge that can negatively affect the performance of these methods. This is because most deep learning models use the entire target image as input. However, but research has shown that with limited training data, the model may not be able to capture discriminative regions of the image. Instead, they might focus on more useless even harmful image regions for recognition, leading to poor recognition results. In this study, we propose a novel SAR ATR framework that addresses the limitation of limited training data. The proposed framework primarily comprises a global assisted branch, locally enhanced branch, feature capture module, and feature discrimination module. The global-assisted branch conducts an initial recognition and provides a loss based on the entire SAR image during each training epoch and the feature capture module automatically segments and captures crucial image regions for current recognition, which we refer to as the “golden key” of the image. The local enhanced branch then performs another recognition and provides another loss function based on the image parts. Instead of updating the model with two basic recognition losses to roughly search for and capture crucial image parts, a feature discrimination module is proposed to combine the global and local branches in a subtle manner to improve local feature separability or compactness for similar inter-class or dissimilar inner-class sample pairs in global features. This adaptively forced the model to capture more crucial image parts and extract more effective features. Experimental results and comparisons of the MSTAR and OpenSARship datasets indicated that the proposed method achieves superior recognition performance compared to existing methods. The effectiveness of our method was further demonstrated through the visualization of the golden key of the testing images and the recognition performance in ablation experiments. We will release our codes and more experimental results at https://github.com/cwwangSARATR/SARATR_FeaCapture_Discrimination.

Medical Image Despeckling Using the Invertible Sparse Fuzzy Wavelet Transform with Nature-Inspired Minibatch Water Wave Swarm Optimization.

Speckle noise is a pervasive problem in medical imaging, and conventional methods for despeckling often lead to loss of edge information due to smoothing. To address this issue, we propose a novel approach that combines a nature-inspired minibatch water wave swarm optimization (NIMWVSO) framework with an invertible sparse fuzzy wavelet transform (ISFWT) in the frequency domain. The ISFWT learns a non-linear redundant transform with a perfect reconstruction property that effectively removes noise while preserving structural and edge information in medical images. The resulting threshold is then used by the NIMWVSO to further reduce multiplicative speckle noise. Our approach was evaluated using the MSTAR dataset, and objective functions were based on two contrasting reference metrics, namely the peak signal-to-noise ratio (PSNR) and the mean structural similarity index metric (MSSIM). Our results show that the suggested approach outperforms modern filters and has significant generalization ability to unknown noise levels, while also being highly interpretable. By providing a new framework for despeckling medical images, our work has the potential to improve the accuracy and reliability of medical imaging diagnosis and treatment planning.

Attention feature fusion awareness network for vehicle target detection in SAR images

ABSTRACT Synthetic aperture radar (SAR) target detection plays a crucial role in military surveillance, earth observation, and disaster monitoring. With the development of deep learning (DL) and SAR imaging technology, numerous SAR target detection methods have been proposed and achieved better detection results. However, detecting different categories of SAR vehicle targets is still challenging due to the influence of coherent speckle noises and background clutter. This article presents a novel attention feature fusion awareness network (AFFNet) for vehicle target detection in SAR images. Specifically, we propose a multi-scale semantic attention (MSSA) module to obtain multi-scale and semantic features of target region; the variable multi-scale feature fusion (VMSFF) module is introduced to effectively fuse different feature information and alleviate target deformation interference by establishing feature correlation; the part feature awareness (PFA) module is used to obtain unique attribute of different vehicle targets to generate accurate anchor boxes. In addition, we design a candidate boundary box selection scheme, which can effectively adapt to SAR targets with different scales and categories. Overall, AFFNet is designed based on the SAR imaging mechanism and target physical feature information. To evaluate the performance of the proposed method, extensive experiments are conducted on the MSTAR dataset. The experiment results show that the proposed AFFNet obtains the mAP of 98.36 and 97.26 on standard operating conditions (SOCs) and extended operating conditions (EOCs), which is more efficient than the other state-of-the-art methods.

Target Recognition in SAR Images Using Complex-Valued Network Guided with Sub-Aperture Decomposition

Synthetic aperture radar (SAR) images have special physical scattering characteristics owing to their unique imaging mechanism. Traditional deep learning algorithms usually extract features from real-valued SAR images in a purely data-driven manner, which may ignore some important physical scattering characteristics and sacrifice some useful target information in SAR images. This undoubtedly limits the improvement in performance for SAR target recognition. To take full advantage of the physical information contained in SAR images, a complex-valued network guided with sub-aperture decomposition (CGS-Net) for SAR target recognition is proposed. According to the fact that different targets have different physical scattering characteristics at different angles, the sub-aperture decomposition is used to improve accuracy with a multi-task learning strategy. Specifically, the proposed method includes main and auxiliary tasks, which can improve the performance of the main task by learning and sharing useful information from the auxiliary task. Here, the main task is the target recognition task, and the auxiliary task is the target reconstruction task. In addition, a complex-valued network is used to extract the features from the original complex-valued SAR images, which effectively utilizes the amplitude and phase information in SAR images. The experimental results obtained using the MSTAR dataset illustrate that the proposed CGS-Net achieved an accuracy of 99.59% (without transfer learning or data augmentation) for the ten-classes targets, which is superior to the other popular deep learning methods. Moreover, the proposed method has a lightweight network structure, which is suitable for SAR target recognition tasks because SAR images usually lack a large number of labeled data. Here, the experimental results obtained using the small dataset further demonstrate the excellent performance of the proposed CGS-Net.

FCCD-SAR: A Lightweight SAR ATR Algorithm Based on FasterNet.

In recent times, the realm of remote sensing has witnessed a remarkable surge in the area of deep learning, specifically in the domain of target recognition within synthetic aperture radar (SAR) images. However, prevailing deep learning models have often placed undue emphasis on network depth and width while disregarding the imperative requirement for a harmonious equilibrium between accuracy and speed. To address this concern, this paper presents FCCD-SAR, a SAR target recognition algorithm based on the lightweight FasterNet network. Initially, a lightweight and SAR-specific feature extraction backbone is meticulously crafted to better align with SAR image data. Subsequently, an agile upsampling operator named CARAFE is introduced, augmenting the extraction of scattering information and fortifying target recognition precision. Moreover, the inclusion of a rapid, lightweight module, denoted as C3-Faster, serves to heighten both recognition accuracy and computational efficiency. Finally, in cognizance of the diverse scales and vast variations exhibited by SAR targets, a detection head employing DyHead's attention mechanism is implemented to adeptly capture feature information across multiple scales, elevating recognition performance on SAR targets. Exhaustive experimentation on the MSTAR dataset unequivocally demonstrates the exceptional prowess of our FCCD-SAR algorithm, boasting a mere 2.72 M parameters and 6.11 G FLOPs, culminating in an awe-inspiring 99.5% mean Average Precision (mAP) and epitomizing its unparalleled proficiency.

An SAR Image Automatic Target Recognition Method Based on the Scattering Parameter Gaussian Mixture Model

General synthetic aperture radar (SAR) image automatic target recognition (ATR) methods perform well under standard operation conditions (SOCs). However, they are not effective in extended operation conditions (EOCs). To improve the robustness of the ATR system under various EOCs, an ATR method for SAR images based on the scattering parameter Gaussian mixture model (GMM) is proposed in this paper. First, an improved active contour model (ACM) is used for target–background segmentation, which is more robust against noise than the constant false alarm rate (CFAR) method. Then, as the extracted attributed scattering center (ASC) is sensitive to noise and resolution, the GMM is constructed using the extracted ASC set. Next, the weighted Gaussian quadratic form distance (WGQFD) is adopted to measure the similarity of GMMs for the recognition task, thereby avoiding false alarms and missed alarms caused by the varying number of scattering centers. Moreover, adaptive aspect–frame division is employed to reduce the number of templates and improve recognition efficiency. Finally, based on the public measured MSTAR dataset, different EOCs are constructed under noise, resolution change, model change, depression angle change, and occlusion of different proportions. The experimental results under different EOCs demonstrate that the proposed method exhibits excellent robustness while maintaining low computation time.

Locality Preserving Property Constrained Contrastive Learning for Object Classification in SAR Imagery

Robust unsupervised feature learning is a critical yet tough task for synthetic aperture radar (SAR) automatic target recognition (ATR) with limited labeled data. The developing contrastive self-supervised learning (CSL) method, which learns informative representations by solving an instance discrimination task, provides a novel method for learning discriminative features from unlabeled SAR images. However, the instance-level contrastive loss can magnify the differences between samples belonging to the same class in the latent feature space. Therefore, CSL can dispel these targets from the same class and affect the downstream classification tasks. In order to address this problem, this paper proposes a novel framework called locality preserving property constrained contrastive learning (LPPCL), which not only learns informative representations of data but also preserves the local similarity property in the latent feature space. In LPPCL, the traditional InfoNCE loss of the CSL models is reformulated in a cross-entropy form where the local similarity of the original data is embedded as pseudo labels. Furthermore, the traditional two-branch CSL architecture is extended to a multi-branch structure, improving the robustness of models trained with limited batch sizes and samples. Finally, the self-attentive pooling module is used to replace the global average pooling layer that is commonly used in most of the standard encoders, which provides an adaptive method for retaining information that benefits downstream tasks during the pooling procedure and significantly improves the performance of the model. Validation and ablation experiments using MSTAR datasets found that the proposed framework outperformed the classic CSL method and achieved state-of-the-art (SOTA) results.

Multi-Aspect SAR Target Recognition Based on Non-Local and Contrastive Learning

Synthetic aperture radar (SAR) automatic target recognition (ATR) has been widely applied in multiple fields. However, the special imaging mechanism of SAR results in different visual features of the same target at different azimuth angles, so single-aspect SAR target recognition has the limitation of observing the target from a single perspective. Multi-aspect SAR target recognition technology can overcome this limitation by utilizing information from different azimuths and effectively improve target recognition performance. Considering the order dependency and data limitation of existing methods, this paper proposes a multi-aspect SAR recognition method based on Non-Local, which applies a self-attention calculation to feature maps to learn the correlation between multi-aspect SAR images. Meanwhile, in order to improve the generalization ability of the proposed method under limited data, a network based on contrastive learning was designed to pre-train the feature extraction part of the whole network. The experimental results using the MSTAR dataset show that the proposed method has excellent recognition accuracy and good robustness.