MSTAR Data Research Articles

The Process Analysis Method of SAR Target Recognition in Pre-Trained CNN Models.

Recently, attention has been paid to the convolutional neural network (CNN) based synthetic aperture radar (SAR) target recognition method. Because of its advantages of automatic feature extraction and the preservation of translation invariance, the recognition accuracies are stronger than traditional methods. However, similar to other deep learning models, CNN is a "black-box" model, whose working process is vague. It is difficult to locate the decision reasons. Because of this, we focus on the process analysis of a pre-trained CNN model. The role of the processing to feature extraction and final recognition decision is discussed. The discussed components of CNN models are convolution, activation function, and full connection. Here, the convolution processing can be deemed as image filtering. The activation function provides a nonlinear element of processing. Moreover, the fully connected layers can also further extract features. In the experiment, four classical CNN models, i.e., AlexNet, VGG16, GoogLeNet, and ResNet-50, are trained by public MSTAR data, which can realize ten-category SAR target recognition. These pre-trained CNN models are processing objects of the proposed process analysis method. After the analysis, the content of the SAR image target features concerned by these pre-trained CNN models is further clarified. In summary, we provide a paradigm to process the analysis of pre-trained CNN models used for SAR target recognition in this paper. To some degree, the adaptability of these models to SAR images is verified.

Distillation Sparsity Training Algorithm for Accelerating Convolutional Neural Networks in Embedded Systems

The rapid development of neural networks has come at the cost of increased computational complexity. Neural networks are both computationally intensive and memory intensive; as such, the minimal energy and computing power of satellites pose a challenge for automatic target recognition (ATR). Knowledge distillation (KD) can distill knowledge from a cumbersome teacher network to a lightweight student network, transferring the essential information learned by the teacher network. Thus, the concept of KD can be used to improve the accuracy of student networks. Even when learning from a teacher network, there is still redundancy in the student network. Traditional networks fix the structure before training, such that training does not improve the situation. This paper proposes a distillation sparsity training (DST) algorithm based on KD and network pruning to address the above limitations. We first improve the accuracy of the student network through KD, and then through network pruning, allowing the student network to learn which connections are essential. DST allows the teacher network to teach the pruned student network directly. The proposed algorithm was tested on the CIFAR-100, MSTAR, and FUSAR-Ship data sets, with a 50% sparsity setting. First, a new loss function for the teacher-pruned student was proposed, and the pruned student network showed a performance close to that of the teacher network. Second, a new sparsity model (uniformity half-pruning UHP) was designed to solve the problem that unstructured pruning does not facilitate the implementation of general-purpose hardware acceleration and storage. Compared with traditional unstructured pruning, UHP can double the speed of neural networks.

A new model based on multi-aspect images and complex-valued neural network for synthetic aperture radar automatic target recognition

ABSTRACT During the last decades, Synthetic Aperture Radar Automatic Target Recognition (SAR ATR) has been vastly utilized in the military services and civil studies. Because of the sensibility of the SAR images of the imaging in the azimuth dimension, using the multi-aspect SAR image sequence is more practical than the single-aspect SAR image to achieve the superior classification accuracy in an actual SAR ATR task. At present time, multi-aspect SAR ATR models mostly utilize Recurrent Neural Networks (RNN) that depend on the sequence between samples and so suffer from the lack of information. In a practical work, a huge amount of training set is needed to train a deep learning network accurately, but it is so costly work to extract multi-aspect SAR images. So, in this article, a new model of multi-aspect SAR ATR is proposed by using a self-attention model. The proposed self-attention model is utilized to calculate the internal correlation between the original SAR images. At the same time, to develop the anti-noise capability of the proposed model and decrease the dependency on a huge amount of training data, a Convolutional Auto Encoder (CAE) is designed and utilized in the feature extraction section of the proposed model. On the other hand, unlike existing traditional methods, in this work we use both amplitude and phase information of SAR images to devolve the training process of the proposed model. It should be noted that all of the parameters of the proposed network are developed to a complex domain. In addition a complex backpropagation algorithm by using the gradient based model is used for training the network. In the end, experiments is obtained by using two MSTAR data set (the MSTAR-SOC and EOC). The experimental results prove that the proposed model not only can achieve a high recognitions rate on the case of sufficient training sample but also can obtain an acceptable rate in the case of small training samples in different complex situations. In addition, the simulation results demonstrate that by using the encoder of CAE in the proposed model, the whole configuration of the proposed model achieves the anti-noise capability that is a valuable benefit of any practical SAR ATR task.

SM-CNN: Separability Measure-Based CNN for SAR Target Recognition

With the maturity of deep learning algorithm in Synthetic Aperture Radar (SAR) target recognition filed, Convolutional Neural Network (CNN) has become the most effective model. However, the interpretability and the separability of feature maps extracted from convolution layers have not been specially analyzed neither qualitatively nor quantitatively, which makes the traditional model work like a “black box”. To alleviate the problem, a novel model based on separability measure (SM) - CNN is proposed in this letter, which introduces the principle of maximal coding rate reduction to the backbone module. SM-CNN quantitatively analyzes the separability of the feature maps and takes the value as a vital part of the loss function to guide the training process of the model. The calculation process of the separability measure values can be strictly derived mathematically, so it is more interpretable, turning the black box into a “gray box”. Additionally, the proposed model can achieve comparable recognition performance of the backbone networks with reduced computational complexity. Comparative experiments based on MSTAR and OpenSARShip data sets verify the effectiveness and practicability of the method proposed in this letter.

PAN: Part Attention Network Integrating Electromagnetic Characteristics for Interpretable SAR Vehicle Target Recognition

Machine learning methods for synthetic aperture radar (SAR) image automatic target recognition (ATR) can be divided into two main types: traditional methods and deep learning methods. The deep learning methods can learn the high-dimensional features of the target directly, and usually obtain high target recognition accuracy. However, they lack full consideration of SAR targets’ inherent characteristics resulting in poor generalization and interpretation ability. Compared with the deep learning methods, traditional methods can get more interpretable and stable results with model-based features. In order to take full advantage of these two kinds of methods, we propose target part attention network based on the attributed scattering center (ASC) model to integrate the electromagnetic characteristics with the deep learning framework. Firstly, considering the importance of scattering structure for SAR ATR, we design a target part model based on ASC model. Then, a novel part attention module based on Scaled Dot-Product Attention mechanism is proposed, which directly associates the features of target parts with the classification results. Finally, we give the derivation method of the importance of each part, which is of great significance for practical application and the interpretation of SAR ATR. Experiments on the MSTAR data set demonstrate the effectiveness of the proposed part attention network. Compared with existing studies, it can achieve higher and more robust classification accuracy under different complex conditions. Furthermore, combined with the importance of parts, we constructed two effective interpretable analysis methods for deep learning network classification results.

Dual Consistency Alignment Based Self-Supervised Learning for SAR Target Recognition With Speckle Noise Resistance

Deep learning based on Convolutional Neural Networks (CNN) has been widely applied in Synthetic Aperture Radar (SAR) target recognition and made significant progress. However, due to the physical effects of the equipment used to collect images, various degrees of speckle noise will be introduced into SAR images. Traditional CNN based SAR target recognition methods are premised on the same noise intensity in the training and testing set, which is contrary to the target recognition in practice. To alleviate this problem, we propose a novel speckle noise resistant framework for SAR target recognition, called Dual Consistency Alignment based Self-Supervised Learning(DCA-SSL). Firstly, original SAR images are randomly added to speckle noise with different thresholds through multiplicative noise, after which contrastive pre-training is performed on unlabeled data. During this period, we combine instance pseudo-label consistency alignment and feature consistency alignment to align multiple threshold speckle noise views with original views under the same targets. Finally, the pre-trained model is migrated to the downstream SAR speckle noise target recognition task. In this article, speckle noise modeling is conducted based on MSTAR data testing set, and experiment results reveal that this method can adapt to different intensities of speckle noise, is robust to modeled SAR image recognition and maintains a high recognition rate even in small sample learning. Our code is available at: <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://github.com/Jordan-Liao/DCA-SSL</uri> .

An Accelerated Hybrid Method for Electromagnetic Scattering of a Composite Target–Ground Model and Its Spotlight SAR Image

In this paper, an accelerated hybrid method of physical optics (PO) shooting and bouncing ray (SBR)–physical theory of diffraction (PTD) is proposed to deal with the electromagnetic scattering of a complex target on rough ground. To accelerate the ray tracing progress, the ray marching technique based on octree structure is employed. In this technique, only the nodes passed by the ray are detected successively until the first facet intersected by ray is found or the ray passes through the bounding box, which greatly decreases the intersection test. Then, based on the accelerated PO-SBR-PTD method, the spotlight synthetic aperture radar (SAR) echo data of the composite target–ground model is obtained by the vector superposition of the echo on each meshed patch. Furthermore, the spotlight SAR image of the composite model is simulated by the polar format algorithm (PFA). In numerical simulations, both the EM scattering of the target and composite model are calculated and evaluated by comparing with the multilevel fast multipole method (MLFMM) in FEKO software. Meanwhile the spotlight SAR image of the composite target–ground model is also compared with the real image in MSTAR data, and a satisfactory similarity between them is obtained. In addition, the SAR images of two targets on rough ground for different pose angles are also presented and analyzed.

Two-dimensional bidirectional principal component collaborative projection feature for SAR vehicle target recognition

With the continuous improvement in the resolution of synthetic aperture radar (SAR), there are many problems in the interpretation of high-resolution SAR images, such as a large amount of data and low efficiency of target recognition. In this paper, a novel SAR target recognition method based on a two-dimensional bidirectional principal component cooperative representation projection feature ((2D)2PCA-CRP) is proposed. First, (2D)2PCA is used to project the image into the low-dimensional feature space, and the redundant information in the high-resolution SAR image is filtered while considering the spatial structure. Then, the spatial global separability feature and local structure feature of the target in the high-resolution SAR image are extracted by CRP to form the (2D)2PCA-CRP feature. Finally, based on this feature, the nearest neighbour classifier is used to complete the target recognition experiments on MSTAR data. The experiments of this study are divided into three parts using standard operation condition (SOC) samples, type change samples and radar incidence angle change data. The experimental results show that the proposed feature achieves better target recognition performance in high-resolution SAR images.

Automated Registration of Multiangle SAR Images Using Artificial Intelligence

Traditionally, nonlinear data processing has been approached via the use of polynomial filters, which are straightforward expansions of many linear methods, or through the use of neural network techniques. In contrast to linear approaches, which often provide algorithms that are simple to apply, nonlinear learning machines such as neural networks demand more computing and are more likely to have nonlinear optimization difficulties, which are more difficult to solve. Kernel methods, a recently developed technology, are strong machine learning approaches that have a less complicated architecture and give a straightforward way to transforming nonlinear optimization issues into convex optimization problems. Typical analytical tasks in kernel-based learning include classification, regression, and clustering, all of which are compromised. For image processing applications, a semisupervised deep learning approach, which is driven by a little amount of labeled data and a large amount of unlabeled data, has shown excellent performance in recent years. For their part, today’s semisupervised learning methods operate on the assumption that both labeled and unlabeled information are distributed in a similar manner, and their performance is mostly impacted by the fact that the two data sets are in a similar state of distribution as well. When there is out-of-class data in unlabeled data, the system’s performance will be adversely affected. When used in real-world applications, the capacity to verify that unlabeled data does not include data that belongs to a different category is difficult to obtain, and this is especially true in the field of synthetic aperture radar image identification (SAR). Using threshold filtering, this work addresses the problem of unlabeled input, including out-of-class data, having a detrimental influence on the performance of the model when it is utilized to train the model in a semisupervised learning environment. When the model is being trained, unlabeled data that does not belong to a category is filtered out by the model using two different sets of data that the model selects in order to optimize its performance. A series of experiments was carried out on the MSTAR data set, and the superiority of our method was shown when it was compared against a large number of current semisupervised classification algorithms of the highest level of sophistication. This was especially true when the unlabeled data had a significant proportion of data that did not fall into any of the categories. The performance of each kernel function is tested independently using two metrics, namely, the false alarm (FA) and the target miss (TM), respectively. These factors are used to calculate the proportion of incorrect judgments made using the techniques.

Dense capsule network for SAR automatic target recognition with limited data

ABSTRACT Automatic target recognition for Synthetic Aperture Radar (SAR) images is a very challengeable problem due to the lack of enough training examples. Currently, there are two options for a network to learn on small SAR data sets: to increase the depth and width or to use the Capsule Network (CapsNet). To take advantage of both options, this paper proposes a Dense CapsNet for SAR target recognition. The contributions are mainly twofold. First, we replace the original convolutional layers with dense blocks in order to increase the depth of the network. Second, we substitute deconvolutional layers for fully connected layers to build the reconstruction network, which is beneficial for improving training efficiency. The experimental results on the MSTAR data set demonstrate that our Dense CapsNet outperforms other state-of-the-art methods upon limited data.

SAR Image Classification Using CNN Embeddings and Metric Learning

The method proposed in this letter for synthetic aperture radar (SAR) image classification has two main stages. In the first stage, a convolutional neural network (CNN) is trained for normal SAR image classification task. After training, the sample features can be obtained by extracting the output of middle layer in the forward propagation process of CNN. In the second stage, an end-to-end metric network is trained to measure the relations between sample features. The method proposed in this letter is tested with some of the larger targets in OpenSARShip data set which is collected from Sentinel-1 satellite, and it is also tested with the MSTAR data set which is created by the U.S. Air Force Laboratory. The experimental results show that our method can get a higher recognition accuracy than normal CNN structure.

SAR image classification based on multi‐feature fusion decision convolutional neural network

With the success of convolutional neural network (CNN) in the field of image processing and image recognition, an improved fully convolutional neural network (FCNN) based on CNN is proposed. Although FCNN has better performance than CNN, the number of parameters of FCNN has also increased significantly compared to CNN. Consequently, this paper proposes a multi-feature fusion decision convolutional neural network (MFFD-CNN), which use two kinds down-sample methods in the pooling layer, and chooses the concatenate or add operation as the way that consolidates the feature of the convolutional layer with a stride of 2 and the max-pooling layer to execute fusion operations. Meanwhile replaces the traditional softmax loss with an additive margin Softmax (AM-Softmax) loss. The structure, which is based on the multi-target classification experiment results of the MSTAR data set, shows can not only obtain an average correct recognition rate 1.2% higher than FCNN, but also better stability without augmenting the training samples of MSTAR data.

Feature Learning for SAR Target Recognition with Unknown Classes by Using CVAE-GAN

Even though deep learning (DL) has achieved excellent results on some public data sets for synthetic aperture radar (SAR) automatic target recognition(ATR), several problems exist at present. One is the lack of transparency and interpretability for most of the existing DL networks. Another is the neglect of unknown target classes which are often present in practice. To solve the above problems, a deep generation as well as recognition model is derived based on Conditional Variational Auto-encoder (CVAE) and Generative Adversarial Network (GAN). A feature space for SAR-ATR is built based on the proposed CVAE-GAN model. By using the feature space, clear SAR images can be generated with given class labels and observation angles. Besides, the feature of the SAR image is continuous in the feature space and can represent some attributes of the target. Furthermore, it is possible to classify the known classes and reject the unknown target classes by using the feature space. Experiments on the MSTAR data set validate the advantages of the proposed method.

Calculated Electronic Energy Loss of Heavy Ions at Low Energies in LR-115, Kapton, SiO2, and Al2O3 Amorphous Materials

The electronic stopping powers of LR-115 and Kapton polymeric foilshave been estimated, using Monte Carlo simulations, for 9 Be, 11 B, 12 C, 14 N, 16 O, and 35 Cl ions covering the energy range ~ 0.1-1.0 MeV/n. Comparison of stopping power based on Lindhard, Scharff, and Schiott (LSS) theory with the corresponding values obtained by SRIM and MSTAR codes in LR-115 and Kapton polymeric foils illustrate a significantly large deviations. However, a semiempirical equation has been proposed here and tested for better stopping power calculations at low-energy regime in the domain of LSS theory for Z = 4-8 ions across materials. Furthermore, the electronic energy losses for 9 Be and 16 O ions in SiO 2 and Al 2 O 3 , respectively, have been calculated in the energy range of ~ 0.1-1.0 MeV/n. The calculated stopping powers exhibit up to 10 % deviation from the experimental values and MSTAR data.

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.

Multiscale Deep Neural Network With Two-Stage Loss for SAR Target Recognition With Small Training Set

Deep learning models have been used recently for target recognition from synthetic aperture radar (SAR) images. However, the performance of these models tends to deteriorate when only a small number of training samples are available due to the problem of overfitting. To address this problem, we propose a two-stage multiscale densely connected convolutional neural networks (TMDC-CNNs). In the proposed TMDC-CNNs, the overfitting issue is addressed with a novel multiscale densely connected network architecture and a two-stage loss function, which integrated the cosine similarity with the prevailing softmax cross-entropy loss. Experiments were conducted on the MSTAR data set, and the results show that our model offers significant recognition accuracy improvements as compared with other state-of-the-art methods, with severely limited training data. The source codes are available at <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://github.com/Stubsx/TMDC-CNNs</uri> .

The Recognition Framework of Deep Kernel Learning for Enclosed Remote Sensing Objects

Remote sensing image target recognition is used in various fields, such as ships, tanks, airplanes, and vehicles, which are closed targets. The features of these targets include target outlines that are obvious and target discriminant features that are significantly different from the surrounding environment, and the targets are characterized as small and dense. Therefore, the recognition of these types of targets is a popular topic. We proposed a recognition framework consisting of a remote sensing image target recognition method based on deep saliency kernel learning analysis, which uses a target region extraction method based on the visual saliency mechanism and implements a nonlinear deep kernel learning saliency feature analysis method to realize target extraction and recognition. Experimental results show that a 95.9% recognition rate is achieved for SAR remote sensing target recognition on the public MSTAR data set, a 96% recognition rate on the UC Merced Land Use data set, and an 85% recognition rate on a self-built visible light remote sensing image data set. The recognition framework can be used for video recognition.

A Novel SAR Image Target Recognition Algorithm under Big Data Analysis

Synthetic aperture radar (SAR) image target recognition technology is aimed at automatically determining the presence or absence of target information from the input SAR image and improving the efficiency and accuracy of SAR image interpretation. Based on big data analysis, dirty data is removed, clean data is returned, and standardized processing of SAR image data is realized. At the same time, by establishing a statistical model of coherent speckles, the convolutional autoencoder is used to denoise the SAR image. Finally, the network model modified by softmax cross‐entropy loss and Fisher loss is used for automatic target recognition. Based on the MSTAR data set, two scene graphs containing the target synthesized by the background image and the target slice are used for experiments. Several comparative experiments have verified the effectiveness of the classification and recognition model in this paper.

Automatic Target Recognition Technology of SAR Images Based on 2DPCA+PNN

In this paper, the SAR image in MSTAR data is used as the research object. The target recognition algorithm based on probabilistic neural network (PNN) is mainly studied. It includes three parts: SAR image preprocessing, feature extraction, classification and recognition. Lee filtering and adaptive threshold method are used to filter the speckle noise effectively, and the 2DPCA principal component analysis method is used to reduce the dimension of the image and obtain the 10 dimensional image features. The recognition part is input to the PNN training test with the acquired feature vectors, and the 85.17% correct recognition rate is obtained, and the target classification and recognition of the SAR image is completed.

Component Interpretation for SAR Target Images Based on Deep Generative Model

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