Automatic Target Recognition Applications Research Articles

Weighted residual network for SAR automatic target recognition with data augmentation.

Decades of research have been dedicated to overcoming the obstacles inherent in synthetic aperture radar (SAR) automatic target recognition (ATR). The rise of deep learning technologies has brought a wave of new possibilities, demonstrating significant progress in the field. However, challenges like the susceptibility of SAR images to noise, the requirement for large-scale training datasets, and the often protracted duration of model training still persist. This paper introduces a novel data augmentation strategy to address these issues. Our method involves the intentional addition and subsequent removal of speckle noise to artificially enlarge the scope of training data through noise perturbation. Furthermore, we propose a modified network architecture named weighted ResNet, which incorporates residual strain controls for enhanced performance. This network is designed to be computationally efficient and to minimize the amount of training data required. Through rigorous experimental analysis, our research confirms that the proposed data augmentation method, when used in conjunction with the weighted ResNet model, significantly reduces the time needed for training. It also improves the SAR ATR capabilities. Compared to existing models and methods tested, the combination of our data augmentation scheme and the weighted ResNet framework achieves higher computational efficiency and better recognition accuracy in SAR ATR applications. This suggests that our approach could be a valuable advancement in the field of SAR image analysis.

Multistage approach for automatic target detection and recognition in infrared imagery using deep learning

Automatic target recognition (ATR) is a challenging task for several computer vision applications. It requires efficient, accurate, and robust methods for target detection and target identification. Deep learning has shown great success in many computer vision applications involving color RGB images. However, the performance of these networks in ATR with infrared sensor data needs further investigation. In this paper, we propose a multistage automatic target detection and recognition (ATDR) system that performs both target detection and target classification on infrared (IR) imagery using deep learning. Our system processes large IR image frames where targets take <1 % of the total number of pixels. First, we train a state-of-the-art object detector you only look once (YOLO) to localize all potential targets in the input image frame. Then, we train a convolutional neural network (CNN) to identify these detections as targets or false alarms. In this second phase, we adapt and analyze the performance of three CNN architectures: a compact and fully connected CNN, VGG16 with batch normalization, and a wide residual neural network (WRN). We also explore the use of a loss function that optimizes directly the area under the receiver operating characteristic (ROC) curve (AUC), and adapt it to our ATR application. To enhance the robustness of the proposed ATR to perturbation and variations introduced during the detection stage, we train our CNN classifiers on automatically detected targets using YOLO, in addition to ground truth bounding boxes and apply selected data augmentation techniques. To simulate real testing environments, where the spatial location of the targets within the image frame is unknown, only YOLO-detected boxes are used during validation. We evaluate our ATDR on a real benchmark dataset that includes different vehicles captured at different resolutions. Our experiments have shown that YOLO can detect most of the targets at the expense of generating a high number of false alarms. We show that the VGG-16 network with batch normalization, which is the best performing model, can correctly identify the classes of the targets, as well as classify the majority of YOLO’s false detections into an additional nontarget class. We also show that the proposed training modification to optimize an AUC-based loss function for ATR proved to be advantageous mainly in identifying difficult targets.

An Empirical Study of Fully Black-Box and Universal Adversarial Attack for SAR Target Recognition

It has been demonstrated that deep neural network (DNN)-based synthetic aperture radar (SAR) automatic target recognition (ATR) techniques are extremely susceptible to adversarial intrusions, that is, malicious SAR images including deliberately generated perturbations that are imperceptible to the human eye but can deflect DNN inference. Attack algorithms in previous studies are based on direct access to a ATR model such as gradients or training data to generate adversarial examples for a target SAR image, which is against the non-cooperative nature of ATR applications. In this article, we establish a fully black-box universal attack (FBUA) framework to craft one single universal adversarial perturbation (UAP) against a wide range of DNN architectures as well as a large fraction of target images. It is of both high practical relevance for an attacker and a risk for ATR systems that the UAP can be designed by an FBUA in advance and without any access to the victim DNN. The proposed FBUA can be decomposed to three main phases: (1) SAR images simulation, (2) substitute model training, and (3) UAP generation. Comprehensive evaluations on the MSTAR and SARSIM datasets demonstrate the efficacy of the FBUA, i.e., can achieve an average fooling ratio of 64.6% on eight cutting-edge DNNs (when the magnitude of the UAP is set to 16/255). Furthermore, we empirically find that the black-box UAP mainly functions by activating spurious features which can effectively couple with clean features to force the ATR models to concentrate on several categories and exhibit a class-wise vulnerability. The proposed FBUA aligns with the non-cooperative nature and reveals the access-free adversarial vulnerability of DNN-based SAR ATR techniques, providing a foundation for future defense against black-box threats.

Physical-Related Feature Extraction From Simulated SAR Image Based on the Adversarial Encoding Network for Data Augmentation

The synthetic aperture radar automatic target recognition (SAR ATR) application based on the deep convolutional network often faces data scarcity. SAR image simulation based on electromagnetic and geometric calculations can provide a large amount of data that contains interpretable physical features, such as shadow and contour. However, there is a big difference between the simulated SAR image and the real image, and it is difficult to directly use it for data augmentation. This letter proposes the adversarial encoding network to extract the physical-related features, which can be understood as the common features between the simulated and real data. By designing the adversarial learning between an encoder and a discriminator, the encoder can extract real features from the simulated images. The encoded features are sent to a classifier to ensure the correct category information. A decoder is used to reconstruct the encoded features into the input image so that the encoded feature retains the image information as much as possible. Ablation experiments and comparative experiments are used to verify the ability of each module and the performance of the proposed method. The results show that the proposed model can achieve 98.55% accuracy, especially when the real data are insufficient for classification, which verifies that the proposed method is effective for data augmentation.

Surrounding Prototype Loss for Radar HRRP Open Set Target Recognition

The open set recognition model can identify the known and unknown samples simultaneously. In the radar automatic target recognition application, open set recognition meets better practical requirements than closed set recognition. Theoretically, an overlap usually exists between the known and unknown features, which makes it difficult for the model to identify unknown samples. Therefore, we explore the distribution of the known and unknown features, and find that the unknown features are usually smaller and closer to the center region in the feature space than the known features. Based on this phenomenon, we propose a novel loss function that improves the open set recognition performance by controlling the known features distributed to the surrounding area of the feature space. In addition, extensive experiments are carried out on measured HRRP data. Thus, we verify the effectiveness of the proposed method.

Scattering Model Guided Adversarial Examples for SAR Target Recognition: Attack and Defense

Deep Neural Networks (DNNs) based Synthetic Aperture Radar (SAR) Automatic Target Recognition (ATR) systems have shown to be highly vulnerable to adversarial perturbations that are deliberately designed yet almost imperceptible but can bias DNN inference when added to targeted objects. This leads to serious safety concerns when applying DNNs to high-stakes SAR ATR applications. Therefore, enhancing the adversarial robustness of DNNs is essential for applying DNNs to modern real-world SAR ATR systems. Toward building more robust DNN-based SAR ATR models, this article explores the domain knowledge of SAR imaging process and proposes a novel Scattering Model Guided Adversarial Attack (SMGAA) algorithm which can generate adversarial perturbations in the form of electromagnetic scattering response (called adversarial scatterers). The proposed SMGAA consists of two parts: 1) a parametric scattering model and corresponding imaging method and 2) a customized gradient-based optimization algorithm. First, we introduce the effective Attributed Scattering Center Model (ASCM) and a general imaging method to describe the scattering behavior of typical geometric structures in the SAR imaging process. By further devising several strategies to take the domain knowledge of SAR target images into account and relax the greedy search procedure, the proposed method does not need to be prudentially finetuned, and can efficiently find the effective ASCM parameters to fool the SAR classifiers and facilitate the robust model training. Comprehensive evaluations on the MSTAR dataset show that the adversarial scatterers generated by SMGAA are more robust to perturbations and transformations in the SAR processing chain than the currently studied attacks, and are effective to construct a defensive model against the malicious scatterers.

A High-Quality Multicategory SAR Images Generation Method With Multiconstraint GAN for ATR

The high-quality training data sets are often insufficient in synthetic aperture radar (SAR) automatic target recognition (ATR) applications. The generative adversarial network (GAN) provides a way for SAR data augmentation. It is necessary to ensure the diversity, similarity, and correct category of the generated images so that these images can be served as the supplementary data set. In this letter, the multiconstraint GAN (MCGAN) is proposed to generate high-quality multicategory SAR images. First, an encoder is used to learn the features of the real images to enhance the similarity. Then, the encoded features are mixed with noise and category labels as the input of the generator to improve the diversity and category correctness. The generated images will be sent to a pretrained classifier to ensure the correct category. Finally, the improved Wasserstein loss with the gradient penalty is extended to the model to further improve the diversity and similarity of the generated images. The MSTAR data set is used to validate the proposed method on generation. The quality evaluation and classification tests are performed on the generated images, and the results show that the MCGAN can provide high-quality images, which could assist in achieving good classification accuracy.

A SAR Image Target Recognition Approach via Novel SSF-Net Models.

With the wide application of high-resolution radar, the application of Radar Automatic Target Recognition (RATR) is increasingly focused on how to quickly and accurately distinguish high-resolution radar targets. Therefore, Synthetic Aperture Radar (SAR) image recognition technology has become one of the research hotspots in this field. Based on the characteristics of SAR images, a Sparse Data Feature Extraction module (SDFE) has been designed, and a new convolutional neural network SSF-Net has been further proposed based on the SDFE module. Meanwhile, in order to improve processing efficiency, the network adopts three methods to classify targets: three Fully Connected (FC) layers, one Fully Connected (FC) layer, and Global Average Pooling (GAP). Among them, the latter two methods have less parameters and computational cost, and they have better real-time performance. The methods were tested on public datasets SAR-SOC and SAR-EOC-1. The experimental results show that the SSF-Net has relatively better robustness and achieves the highest recognition accuracy of 99.55% and 99.50% on SAR-SOC and SAR-EOC-1, respectively, which is 1% higher than the comparison methods on SAR-EOC-1.

Inherent limitation of digital imagery: spatial-phase vacillations and the ambiguity function

The spatial relationship of a scene image upon a pixilated focal-plane-array can temporally change due to scene or camera motion, microjitter of camera line-of-sight, etc. It was found that degradation caused by subpixel spatial-phase vacillations (SPV) upon the performance of matched filters was unexpectedly significant. Subpixel spatial-phase vacillations can cause degradations in matched filter performance to detect desired objects and discriminate other objects. SPV appear to be an inherent limitation of digital imagery when processed using matched filter methodology and can negatively impact the performance of systems. Mitigation of this degradation was found to be possible by utilizing one of several matched filter constructions such as a multi-filter enhanced matched filter (EMF) bank and a single-filter EMF. A significant conclusion of this investigation is that, for automatic target recognition applications, improved overall task performance should be realized by the use of an EMF bank. Dramatic reduction in computational resources for image matching multispectral imagery was accomplished by spectrally-collapsing multispectral imagery to form pseudoimages with the spectral information appearing as a texture in a grayscale image. The pattern recognition ambiguity function, which sets the fundamental performance limit of an image processing system, is introduced.

Class Boundary Exemplar Selection Based Incremental Learning for Automatic Target Recognition

When adding new tasks/classes in an incremental learning scenario, the previous recognition capabilities trained on the previous training data can be lost. In the real-life application of automatic target recognition (ATR), part of the previous samples may be able to be used. Most incremental learning methods have not considered how to save the previous key samples. In this article, the class boundary exemplar selection-based incremental learning (CBesIL) is proposed to save the previous recognition capabilities in the form of the class boundary exemplars. For exemplar selection, the class boundary selection method based on local geometrical and statistical information is proposed. And when adding new classes continually, a class-boundary-based data reconstruction method is introduced to update the exemplar set. Thus, when adding new classes, the previous class boundaries could be kept complete. Experimental results demonstrate that the proposed CBesIL outperforms the other state of the art on the accuracy of multiclass recognition and class-incremental recognition.

Improving SAR Automatic Target Recognition Models With Transfer Learning From Simulated Data

Data-driven classification algorithms have proved to do well for automatic target recognition (ATR) in synthetic aperture radar (SAR) data. Collecting data sets suitable for these algorithms is a challenge in itself as it is difficult and expensive. Due to the lack of labeled data sets with real SAR images of sufficient size, simulated data play a big role in SAR ATR development, but the transferability of knowledge learned on simulated data to real data remains to be studied further. In this letter, we show the first study of Transfer Learning between a simulated data set and a set of real SAR images. The simulated data set is obtained by adding a simulated object radar reflectivity to a terrain model of individual point scatters, prior to focusing. Our results show that a Convolutional Neural Network (Convnet) pretrained on simulated data has a great advantage over a Convnet trained only on real data, especially when real data are sparse. The advantages of pretraining the models on simulated data show both in terms of faster convergence during the training phase and on the end accuracy when benchmarked on the Moving and Stationary Target Acquisition and Recognition data set. These results encourage SAR ATR development to continue the improvement of simulated data sets of greater size and complex scenarios in order to build robust algorithms for real life SAR ATR applications.

Infrared variation reduction by simultaneous background suppression and target contrast enhancement for deep convolutional neural network-based automatic target recognition

Automatic target recognition (ATR) is a traditionally challenging problem in military applications because of the wide range of infrared (IR) image variations and the limited number of training images. IR variations are caused by various three-dimensional target poses, noncooperative weather conditions (fog and rain), and difficult target acquisition environments. Recently, deep convolutional neural network-based approaches for RGB images (RGB-CNN) showed breakthrough performance in computer vision problems, such as object detection and classification. The direct use of RGB-CNN to the IR ATR problem fails to work because of the IR database problems (limited database size and IR image variations). An IR variation-reduced deep CNN (IVR-CNN) to cope with the problems is presented. The problem of limited IR database size is solved by a commercial thermal simulator (OKTAL-SE). The second problem of IR variations is mitigated by the proposed shifted ramp function-based intensity transformation. This can suppress the background and enhance the target contrast simultaneously. The experimental results on the synthesized IR images generated by the thermal simulator (OKTAL-SE) validated the feasibility of IVR-CNN for military ATR applications.

3D Imaging of Rapidly Spinning Space Targets Based on a Factorization Method.

Three-dimensional (3D) imaging of space targets can provide crucial information about the target shape and size, which are significant supports for the application of automatic target classification and recognition. In this paper, a new 3D imaging of space spinning targets via a factorization method is proposed. Firstly, after the translational compensation, the scattering centers two-dimensional (2D) range and range-rate sequence induced by the target spinning is extracted using a high resolution spectral estimation technique. Secondly, measurement data association is implemented to obtain the scattering center trajectory matrix by using a range-Doppler tracker. Then, we use an initial coarse angular velocity to generate the projection matrix, which consists of the scattering centers range and cross-range, and a factorization method is applied iteratively to the projection matrix to estimate the accurate angular velocity. Finally, we use the accurate estimate spinning angular velocity to rescale the projection matrix and the well-scaled target 3D geometry is reconstructed. Compared to the previous literature methods, ambiguity in the spatial axes can be removed by this method. Simulation results have demonstrated the effectiveness and robustness of the proposed method.

Feature Fusion Based on Convolutional Neural Network for SAR ATR

Recent breakthroughs in algorithms related to deep convolutional neural networks (DCNN) have stimulated the development of various of signal processing approaches, where the specific application of Automatic Target Recognition (ATR) using Synthetic Aperture Radar (SAR) data has spurred widely attention. Inspired by the more efficient distributed training such as inception architecture and residual network, a new feature fusion structure which jointly exploits all the merits of each version is proposed to reduce the data dimensions and the complexity of computation. The detailed procedure presented in this paper consists of the fused features, which make the representation of SAR images more distinguishable after the extraction of a set of features from DCNN, followed by a trainable classifier. In particular, the obtained results on the 10-class benchmark data set demonstrate that the presented architecture can achieve remarkable classification performance to the current state-of-the-art methods.

Multi‐scale feature‐based fuzzy‐support vector machine classification using radar range profiles

High resolution range profile possesses great significance in automatic target recognition (ATR) research and application due to its efficiency superiority. Feature extraction is one of the major ATR research directions. Existing feature extraction techniques only concern the range profile in original scale, which constrains the cognitive level of classifiers. This study goes beyond this limitation and introduces scale space theory in the feature extraction of range profiles. The edge points of range profiles are extracted and a novel multi-scale target classification method based on fuzzy-support vector machine classifier is proposed. Range profiles of three aircraft models are selected to validate the classification method. Experiment results demonstrate that the addition of edge points is favourable to enhance range profile samples' separability. Comparison of classification performance between original and multiple scales validates multi-scale technique's feasibility in range profile classification problem. The weighting factor developed according to the quadratic entropy optimises range profile samples' membership degree distribution and brings further promotion on classification performance.

Analgorithmic Framework for Automatic Detection and Tracking Moving Point Targets in IR Image Sequences

Imaging sensors operating in infrared (IR) region of electromagnetic spectrum are gaining importance in airborne automatic target recognition (ATR) applications due to their passive nature of operation. IR imaging sensors exploit the unintended IR radiation emitted by the targets of interest for detection. The ATR systems based on the passive IR imaging sensors employ a set of signal processing algorithms for processing the image information in real-time. The real-time execution of signal processing algorithms provides the sufficient reaction time to the platform carrying ATR system to react upon the target of interest. These set of algorithms include detection, tracking, and classification of low-contrast, small sized-targets. Paper explained a signal processing framework developed to detect and track moving point targets from the acquired IR image sequences in real-time. Defence Science Journal, Vol. 65, No. 3, May 2015, pp.208-213, DOI: http://dx.doi.org/10.14429/dsj.65.8164

Fast image matching algorithm based on affine invariants

Feature-based image matching algorithms play an indispensable role in automatic target recognition (ATR). In this work, a fast image matching algorithm (FIMA) is proposed which utilizes the geometry feature of extended centroid (EC) to build affine invariants. Based on affine invariants of the length ratio of two parallel line segments, FIMA overcomes the invalidation problem of the state-of-the-art algorithms based on affine geometry features, and increases the feature diversity of different targets, thus reducing misjudgment rate during recognizing targets. However, it is found that FIMA suffers from the parallelogram contour problem and the coincidence invalidation. An advanced FIMA is designed to cope with these problems. Experiments prove that the proposed algorithms have better robustness for Gaussian noise, gray-scale change, contrast change, illumination and small three-dimensional rotation. Compared with the latest fast image matching algorithms based on geometry features, FIMA reaches the speedup of approximate 1.75 times. Thus, FIMA would be more suitable for actual ATR applications.

Automatic target recognition of synthetic aperture radar (SAR) images based on optimal selection of Zernike moments features

In the present study, a new algorithm for automatic target detection (ATR) in synthetic aperture radar (SAR) images has been proposed. First, moving and stationary target acquisition and recognition image chips have been segmented and then passed to a number of preprocessing stages such as histogram equalisation, position and size normalisation. Second, the feature extraction based on Zernike moments (ZMs) having linear transformation invariance properties and robustness in the presence of the noise has been introduced for the first time. Third, a genetic algorithm‐based feature selection and a support vector machine classifier have been presented to select the optimal feature subset of ZMs for decreasing the computational complexity. Experimental results demonstrate the efficiency of the proposed approach in target recognition of SAR imagery. The authors obtained results show that just a small amount of ZMs features is sufficient to achieve the recognition rates that rival other established methods, and so ZMs features can be regarded as a powerful discriminatory feature for automatic target recognition applications relevant to SAR imagery. Furthermore, it can be observed that the classifier performs fairly well until the signal‐to‐noise ratio falls beneath 5 dB for noisy images.

Robust Target Identification Using a Modified Generalized Likelihood Ratio Test

In order to correctly identify a remote target, an efficient and robust target signature identification technique is required. Radar target identification based on complex natural resonances (CNRs) has drawn the interest of many researchers following the development of the singularity expansion method (SEM). As evident from the literature, statistical techniques such as the generalized likelihood ratio test (GLRT) have produced a better identification result, in the presence of noise, compared to some other SEM-based identification methods such as the extinction pulse (E-pulse) technique. However, one of the issues related to a resonance based target classifier is that it requires the commencement of the late time period for the unknown target response to be determined accurately in order to avoid false alarms during target classification process. For automatic target recognition (ATR) applications, usually such information is not known a priori. In view of this problem, a modified GLRT technique that utilizes time-frequency analysis is presented in this paper. The improved GLRT method does not require prior knowledge of the beginning of the late time period for the transient response of the unknown target. Simulation results using various targets show that our method is comparable to the original GLRT technique when the commencement of the late time period for the unknown target response is correctly determined and outperforming the original GLRT technique when the commencement of the late time period for the unknown target response is incorrectly determined.

Inverse synthetic aperture radar image focusing using fast adaptive joint time–frequency and three-dimensional motion detection on experimental radar data

This study presents two algorithms for enhancing the capabilities of the adaptive joint time-frequency (AJTF) approach and the three-dimensional (3D) motion detection method, respectively, on inverse synthetic aperture radar (ISAR) imaging. The first method is a modified AJTF motion compensation algorithm that is able to employ more than two prominent point scatterers for translation and rotation motion compensation, respectively. The second algorithm enhances a current 3D motion detection method. This algorithm improves the computation time for identifying 3D motion in ISAR images and selects the imaging time intervals with the least 3D motion. With these enhanced detection algorithms, the authors have the ability to distinguish the time intervals when the target undergoes smooth two-dimensional (2D) motion from those containing more chaotic 3D motion. As a result, the authors can reliably detect those time intervals where 2D target motions are predominant to form well-focused ISAR images that are used in automatic target recognition applications.