Abstract: This paper presents an approach for attributed scattering center (ASC) matching with application to synthetic aperture radar (SAR) automatic target recognition (ATR). A statistics-based distance measure is designed to evaluate the distance between individual ASCs. Afterwards, the Hungarian algorithm is employed to build a one-to-one correspondence between two ASC sets. Based on the correspondence, a global similarity and a local similarity are designed to comprehensively evaluate the global consistency and structural correlation between those two ASC sets. The two similarities comprehensively exploit the inner correlation between the two ASC sets, thus providing a reliable and robust similarity measure for SAR ATR. The two similarities are then fused based on the Dempster–Shafer evidence theory to determine the target type by the maximum belief rule. Extensive experiments conducted on the moving and stationary target acquisition and recognition dataset and the comparison with several state-of-the-art methods demonstrate the validity and robustness of the proposed method.

Abstract: Synthetic aperture radar (SAR) image simulators based on computer-aided drawing models play an important role in SAR applications, such as automatic target recognition and image interpretation. However, the accuracy of such simulators is due to geometric error and simplification in the electromagnetic calculation. In this letter, an end-to-end model was developed that could directly synthesize the desired images from the known image database. The model was based on generative adversarial nets (GANs), and its feasibility was validated by comparisons with real images and ray-tracing results. As a further step, the samples were synthesized at angles outside of the data set. However, the training process of GAN models was difficult, especially for SAR images which are usually affected by noise interference. The major failure modes were analyzed in experiments, and a clutter normalization method was proposed to ameliorate them. The results showed that the method improved the speed of convergence up to 10 times. The quality of the synthesized images was also improved.

Abstract: Deep learning algorithms have been introduced into target recognition of synthetic aperture radar (SAR) images for extracting deep features because of its accuracy on various recognition problems with sufficient training samples. However, applying deep structures in recognizing SAR images may suffer lack of training samples. Therefore, a deep learning method is proposed in this study based on a multilayer autoencoder (AE) combined with a supervised constraint. We bind the original AE algorithm with a restriction based on Euclidean distance to use the limited training images well. Moreover, a dropout step is added to our algorithm, which is designed to prevent overfitting caused by supervised learning. Experimental results on the MSTAR dataset demonstrate the effectiveness of the proposed method on real SAR images.

Abstract: This paper presents an automatic target recognition (ATR) approach for sonar onboard unmanned underwater vehicles (UUVs). In this approach, target features are extracted by a convolutional neural network (CNN) operating on sonar images, and then classified by a support vector machine (SMV) that is trained based on manually labeled data. The proposed approach is tested on a set of sonar images obtained by a UUV equipped with side-scan sonar. Automatic target recognition is achieved through the use of matched filters, while target classification is achieved with the trained SVM classifier based on features extracted by the CNN. The results show that deep learning feature extraction provide better performance compared to using other feature extraction techniques such as histogram of oriented gradients (HOG) and local binary pattern (LBP). By processing images autonomously, the proposed approach can be combined with onboard planning and control systems to develop autonomous UUVs able to search for underwater targets without human intervention.

Abstract: Motivated by the recent advances in deep learning, we lay out a vision of how deep learning techniques can be used in radar. Specifically, our discussion focuses on the use of deep learning to advance the state-of-the-art in radar imaging. While deep learning can be directly applied to automatic target recognition (ATR), the relevance of these techniques in other radar problems is not obvious. We argue that deep learning can play a central role in advancing the state-of-the-art in a wide range of radar imaging problems, discuss the challenges associated with applying these methods, and the potential advancements that are expected. We lay out an approach to design a network architecture based on the specific structure of the synthetic aperture radar (SAR) imaging problem that augments learning with traditional SAR modelling. This framework allows for capture of the non-linearity of the SAR forward model. Furthermore, we demonstrate how this process can be used to learn and compensate for trajectory based phase error for the autofocus problem.

Abstract: Feature extraction has always been a difficult problem in the classification performance of synthetic aperture radar automatic target recognition (SAR-ATR). It is very important to select discriminative features to train a classifier, which is a prerequisite. Inspired by the great success of convolutional neural network (CNN), we address the problem of SAR target classification by proposing a feature extraction method, which takes advantage of exploiting the extracted deep features from CNNs on SAR images to introduce more powerful discriminative features and robust representation ability for them. First, the pretrained VGG-S net is fine-tuned on moving and stationary target acquisition and recognition (MSTAR) public release database. Second, after a simple preprocessing is performed, the fine-tuned network is used as a fixed feature extractor to extract deep features from the processed SAR images. Third, the extracted deep features are fused by using a traditional concatenation and a discriminant correlation analysis algorithm. Finally, for target classification, K-nearest neighbors algorithm based on LogDet divergence-based metric learning triplet constraints is adopted as a baseline classifier. Experiments on MSTAR are conducted, and the classification accuracy results demonstrate that the proposed method outperforms the state-of-the-art methods.

Abstract: The challenge of automatic target recognition of military targets within a synthetic aperture radar scene is addressed in this paper. The proposed approach exploits the discrete-defined Krawtchouk moments, which are able to represent a detected extended target with few features, allowing its characterization. The proposed algorithm provides robust performance for target recognition, identification, and characterization, with high reliability in the presence of noise and reduced sensitivity to discretization errors. The effectiveness of the proposed approach is demonstrated using the MSTAR dataset.

Abstract: The quality of synthetic aperture radar (SAR) images and the completeness of the template database are two important factors in template-based SAR automatic target recognition. This letter gives a solution to the two factors by multilevel reconstruction of SAR targets using attributed scattering centers (ASCs). The ASCs of original SAR images are extracted to reconstruct the target’s image, which not only reduces the noise and background clutters but also keeps the electromagnetic characteristics of the target. Template database are reconstructed at multilevels to simulate various extents of ASC absence in the extended operation conditions. Therefore, the quality of SAR images as well as the completeness of the template database is augmented. Features are extracted from the augmented SAR images, and the classifier is trained by the augmented database for target recognition. Experimental results on the moving and stationary target acquisition and recognition data set demonstrate the validity of the proposed method.

Abstract: The outstanding pattern recognition performance of deep learning brings new vitality to the synthetic aperture radar (SAR) automatic target recognition (ATR). However, there is a limitation in current deep learning based ATR solution that each learning process only handles one SAR image, namely learning the static scattering information, while missing the space-varying information. It is obvious that space-varying scattering information introduced in the multi-aspect joint recognition should improve the classification accuracy and robustness. In this paper, a novel multi-aspect-aware method is proposed to achieve this idea through the bidirectional long short-term memory (LSTM) recurrent neural networks-based space-varying scattering information learning. Specifically, we first select different aspect images to generate the multi-aspect space-varying image sequences. Then, the Gabor filter and three-patch local binary pattern are progressively implemented to extract comprehensive spatial features, followed by dimensionality reduction with the multi-layer perceptron network. Finally, we design a bidirectional LSTM recurrent neural network to learn the multi-aspect features with further integrating the softmax classifier to achieve target recognition. Experimental results demonstrate that the proposed method can achieve 99.9% accuracy for 10-class recognition. Besides, its anti-noise and anti-confusion performances are also better than the conventional deep learning-based methods.

Abstract: Automatic target recognition has been widely studied over the years, yet it is still an open problem. The main obstacle consists in extended operating conditions, e.g. ., depression angle change, configuration variation, articulation, and occlusion. To deal with them, this paper proposes a new classification strategy. We develop a new representation model via the steerable wavelet frames. The proposed representation model is entirely viewed as an element on Grassmann manifolds. To achieve target classification, we embed Grassmann manifolds into an implicit reproducing Kernel Hilbert space (RKHS), where the kernel sparse learning can be applied. Specifically, the mappings of training sample in RKHS are concatenated to form an overcomplete dictionary. It is then used to encode the counterpart of query as a linear combination of its atoms. By designed Grassmann kernel function, it is capable to obtain the sparse representation, from which the inference can be reached. The novelty of this paper comes from: 1) the development of representation model by the set of directional components of Riesz transform; 2) the quantitative measure of similarity for proposed representation model by Grassmann metric; and 3) the generation of global kernel function by Grassmann kernel. Extensive comparative studies are performed to demonstrate the advantage of proposed strategy.

Abstract: The frequency of terahertz radar ranges from 0.1 THz to 10 THz, which is higher than that of microwaves. Multi-modal signals, including high-resolution range profile (HRRP) and Doppler signatures, can be acquired by the terahertz radar system. These two kinds of information are commonly used in automatic target recognition; however, dynamic gesture recognition is rarely discussed in the terahertz regime. In this paper, a dynamic gesture recognition system using a terahertz radar is proposed, based on multi-modal signals. The HRRP sequences and Doppler signatures were first achieved from the radar echoes. Considering the electromagnetic scattering characteristics, a feature extraction model is designed using location parameter estimation of scattering centers. Dynamic Time Warping (DTW) extended to multi-modal signals is used to accomplish the classifications. Ten types of gesture signals, collected from a terahertz radar, are applied to validate the analysis and the recognition system. The results of the experiment indicate that the recognition rate reaches more than 91%. This research verifies the potential applications of dynamic gesture recognition using a terahertz radar.

Abstract: The detection performance of radar maneuvering target with jerk motion is affected by the range migration (RM) and Doppler frequency migration (DFM). To address these problems, a fast algorithm without searching target's motion parameters is proposed. In this algorithm, the second-order keystone transform is first applied to eliminate the quadratic coupling between the range frequency and slow time. Then, by employing a new defined symmetric autocorrelation function, scaled Fourier transform, and inverse fast Fourier transform, the target's initial range and velocity are estimated. With these two estimates, the azimuth echoes along the target's trajectory, which can be modeled as a cubic phase signal (CPS), are extracted. Thereafter, the target's radial acceleration and jerk are estimated by approaches for parameters estimation of the CPS. Finally, by constructing a compensation function, the RM and DFM are compensated simultaneously, followed by the coherent integration and target detection. Comparisons with other representative algorithms in computational cost, motion parameter estimation performance, and detection ability indicate that the proposed algorithm can achieve a good balance between the computational cost and detection ability. The simulation and raw data processing results demonstrate the effectiveness of the proposed algorithm.

Abstract: This paper deals with the problematic of automatic target recognition (ATR) using Synthetic Aperture Radar (SAR) images. In this work, the Deep Learning (DL) architecture is proposed and applied in order to recognize military vehicles from SAR images. We propose mainly in this work the deep learning algorithms based on convolutional neural network architecture. In the second step and in order to optimize the convolution of DL steps, we propose to use a convolutional auto-encoder which may be better suited to image processing. Its use provides several areas of the best results in the presence of noise on shifted and truncated images. To validate our approach, some experimentation results are given and compared. The obtained results show that the proposed approach of DL achieves a height recognition accuracy of 93%.

Abstract: This report deals with translation invariance of convolutional neural networks (CNNs) for automatic target recognition (ATR) from synthetic aperture radar (SAR) imagery. In particular, the translation invariance of CNNs for SAR ATR represents the robustness against misalignment of target chips extracted from SAR images. To understand the translation invariance of the CNNs, we trained CNNs which classify the target chips from the MSTAR into the ten classes under the condition of with and without data augmentation, and then visualized the translation invariance of the CNNs. According to our results, even if we use a deep residual network, the translation invariance of the CNN without data augmentation using the aligned images such as the MSTAR target chips is not so large. A more important factor of translation invariance is the use of augmented training data. Furthermore, our CNN using augmented training data achieved a state-of-the-art classification accuracy of 99.6%. These results show an importance of domain-specific data augmentation.

Abstract: Zero-shot learning (ZSL) is of critical importance for practical synthetic aperture radar (SAR) automatic target recognition (ATR) as training samples are not always available for all targets and all observation configurations. We propose a novel generative-based deep neural network framework for ZSL of SAR ATR. The key component of the framework is a generative deconvolutional neural network referred to as generator. It learns a faithful hierarchical representation of known targets while automatically constructing a continuous SAR target feature space spanned by orientation-invariant features and orientation angle. It is then used as a reference to design and initialize an interpreter convolutional neural network, which is inversely symmetric to the generator network. The interpreter network is then trained to map any input SAR image, including those of unseen targets, into the target feature space. In a preliminary experiment with the Moving and Stationary Target Acquisition and Recognition data set, seven targets are used in the training of generator and interpreter networks. Then, the eighth target is used to test the interpreter, where it is correctly mapped to the reasonable spot spanned by the previous seven targets and its orientation can also be estimated.

Abstract: The Sparse Fourier Transform (SFT), designed for signals that contain a small number of frequencies, enjoys low complexity, and thus is ideally suited for big data applications. In this paper, we propose Robust Sparse Fourier Transform, (RSFT), which is a modification of SFT that extends the SFT advantages to real world, noisy settings. RSFT can accommodate off-grid frequencies in the data. Furthermore, by incorporating Neyman–Pearson detection in the SFT stages, frequency detection does not require knowledge of the exact sparsity of the signal, and is robust to noise. We analyze the asymptotic performance of RSFT, and study the computational complexity versus detection performance tradeoff. We show that, by appropriately choosing the detection thresholds, the optimal tradeoff can be achieved. We discuss the application of RSFT on short-range ubiquitous radar signal processing and demonstrate its feasibility via simulations.

Abstract: Multi-feature decision fusion is a robust way to enhance the performance of synthetic aperture radar (SAR) automatic target recognition (ATR). This study fuses two physically relevant features namely attributed scattering centres and target contour for SAR ATR. The two features provide an intuitive and comprehensive description of targets which can not only promote the ATR performance but also benefit human interpretation. The two features complement each other with weak spatial correlation. Furthermore, both the features can be concisely represented by simple point pattern thus remarkably saving the memory of the feature database. Specific classifiers are designed for the two features whose outputs are fused by the Dempster–Shafer evidence theory to produce the final decision. Experiments based on the moving and stationary target acquisition and recognition database demonstrate the validity of the proposed method.

Abstract: Display Omitted A new bee colony in each generation is introduced to avoid high levels of inbreeding.A population initialization strategy is adopted to make the search more efficient and faster.Bayes theorem is applied to adjust the breeding coefficient in real time to accelerate the algorithm convergence speed. The image thresholding techniques are considered as a must for objects segmentation, compression and target recognition, and they have been widely studied for the last few decades; for example, the multi-level thresholding methods, and as such (they) render more great challenges for image segmentation techniques that remain computationally more expensive, when their choices of threshold numbers were increased. Therefore, our aim was to propose an algorithm based on Bayesian theorem and the so-called honey-bee-mating algorithm (HBMA), called a Bayesian honey bee mating algorithm BHBMA. It can not only reduce the computational time and curse of dimensionality, but also can run more reliably and more stably. This enhanced capability was technically accomplished by embedding a new population initialization strategy based on the characteristics of multi-level thresholding technique in pixel-based intensity images arranged from lower grey levels to higher ones. Extensive experiments have shown that our proposed method outperformed other state-of-the-art algorithms empirically in terms of their effectiveness and efficiency, when applying to complex image processing scenario such as automatic target recognition.

Abstract: In this study, we suggest additional feature based convolutional neural networks (CNN) for synthetic aperture radar automatic target recognition (SAR ATR) performance improvement. Previous SAR ATR researches need preprocessing process or prior information such as pose information due to severe image noise on SAR image. However, since the noise characteristics of the SAR image are different for each acquired image, recognition accuracy may be lowered if the preprocessing process is not performed properly. For this reason, we propose additional feature based CNN architecture which does not need additional preprocessing process or pose information. The proposed method consists of three steps. First, extract more detail information included features and noise reduced features from two CNNs using max-pool and average-pool subsampling operation. Second, the features extracted from the two CNNs are aggregated into a single column vector in order to consider both features in target recognition. Lastly, train proposed CNN architecture using aggregated features and fully-connected layers. MSTAR SAR dataset is used for simulation and confirmed that proposed method can recognize targets more accurately than other methods. Using the proposed method in this study, we can recognize the 10 classes of military targets with accuracy of 94.38% without any additional preprocess or prior information.

Abstract: Inverse synthetic aperture radar (ISAR) cross-range scaling is used to obtain the actual cross-range size of the target, which is essential for space surveillance and automatic target recognition. In this letter, a novel two-step ISAR cross-range scaling method for earth-orbit targets is proposed, which improves the computational efficiency through the use of prior information and achieves high estimation accuracy. An initial rotation velocity (RV) is calculated first using the open two-line element data of the satellite orbit to coarsely achieve a cross-range scaling of the ISAR image with high efficiency. Then, the refined cross-range scaling result is obtained with an accurate RV, which is accomplished by the isolated scatterer extraction and the chirp-rate estimation, wherein the blob detection and the integrated cubic phase function are employed, respectively. The initial RV is used to narrow the search width of the chirp-rate estimation, and the corresponding computational burden is expected to decrease accordingly. Finally, simulations and real-data experiments are performed to verify the effectiveness and the accuracy of the proposed method.

Abstract: In this article, a new type of feature, named two-dimensional 2D-slice Zernike moments, is proposed for synthetic aperture radar SAR automatic target recognition ATR. Target features play an extremely important role in the ATR system. Pixels with different scattering intensities distribute in different positions in SAR images, which represent target inherent signatures determined by the target’s characteristics, including global structure and local details. To extract these various scattering signatures, we developed a feature extraction technique named 2D-slice Zernike moments 2DS-ZMS, which can capture target global and local scattering field distribution information. First, the multilayer 2D-slices of a SAR image are extracted by uniformly cutting the 3D display SAR image along the amplitude direction. Then Zernike moments of each 2D-slice are calculated. Finally, the 2DS-ZMS of the SAR image are formulated into a column vector, called the feature vector. The obtained feature vectors of the targets are fed into a newly developed classifier, i.e. the sparse representation-based classifier SRC. By projecting the testing sample feature vector on the dictionary made up of training samples feature vectors, the sparse coefficients are solved. The minimum reconstruction residual is adopted as the judgement criterion for predicting the test sample’s class label. Experiments on the moving and stationary target acquisition and recognition MSTAR data set validate the effectiveness and superiority of the proposed method.

Abstract: SAR is a very popular instrument for imaging of the ground surface. Possibility of high-resolution image formation makes it superior tool for various information extraction tasks. In the paper, a problem of automatic target recognition is comprehensively analyzed. An idea of azimuth and range target profiles fusion is proposed. It is demonstrated, that usage of a proper image preprocessing with appropriate feature extraction steps allow to achieve a competitive recognition accuracy while keeping a low-dimensionality of feature vectors. Experimental results are discussed for a publicly available MSTAR dataset.

Abstract: Future light detection and ranging seeker missiles incorporating 3D automatic target recognition (ATR) capabilities can improve the missile’s effectiveness in complex battlefield environments. Considering the progress of local 3D descriptors in the computer vision domain, this paper evaluates a number of these on highly credible simulated air-to-ground missile engagement scenarios. The latter take into account numerous parameters that have not been investigated yet by the literature including variable missile – target range, 6-degrees-of-freedom missile motion and atmospheric disturbances. Additionally, the evaluation process utilizes our suggested 3D ATR architecture that compared to current pipelines involves more post-processing layers aiming at further enhancing 3D ATR performance. Our trials reveal that computer vision algorithms are appealing for missile-oriented 3D ATR.

Abstract: This study reports on a novel technique for the estimation of the target total rotational velocity of a non-cooperative target. The technique makes use of the second-order local polynomial Fourier transform to estimate the hitherto unknown component of the velocity vector in the radar line-of-sight direction, and three-dimensional (3D) interferometric inverse synthetic aperture radar (ISAR) imaging the target. The aspect-independent 3D size and shape of a target – an important metric for automatic target recognition – can thus be estimated. Cross-range focus in the component 2D ISAR images is also improved in the process. Results obtained from extensive simulations demonstrate the feasibility of the proposed approach.

Abstract: Fast, accurate and robust automatic target recognition (ATR) in optical aerial imagery can provide game-changing advantages to military commanders and personnel. ATR algorithms must reject non-targets with a high degree of confidence in a world with an infinite number of possible input images. Furthermore, they must learn to recognize new targets without requiring massive data collections. Whereas most machine learning algorithms classify data in a closed set manner by mapping inputs to a fixed set of training classes, open set recognizers incorporate constraints that allow for inputs to be labelled as unknown. We have adapted two template-based open set recognizers to use computer generated synthetic images of military aircraft as training data, to provide a baseline for military-grade ATR: (1) a frequentist approach based on probabilistic fusion of extracted image features, and (2) an open set extension to the one-class support vector machine (SVM). These algorithms both use histograms of oriented gradients (HOG) as features as well as artificial augmentation of both real and synthetic image chips to take advantage of minimal training data. Our results show that open set recognizers trained with synthetic data and tested with real data can successfully discriminate real target inputs from non-targets. However, there is still a requirement for some knowledge of the real target in order to calibrate the relationship between synthetic template and target score distributions. We conclude by proposing algorithm modifications that may improve the ability of synthetic data to represent real data.

Abstract: In this letter, we propose a novel classification strategy called the coupled dictionary learning for target recognition in synthetic aperture radar (SAR) images. First, we train structured synthesis dictionaries to reflect the difference among each category. Second, we introduce a shared dictionary to reduce the effect of common features, such as the high similarity caused by specular reflection. Finally, we use the analysis dictionary to improve the efficiency of recognition by eliminating the constraint of the $l_{0}$ -norm or $l_{1}$ -norm of sparse code. Experimental results on Moving and Stationary Target Acquisition and Recognition data set indicate that our method can achieve better performance in SAR target recognition than the state-of-the-art methods, such as tritask joint sparse representation and CKLR. Especially, this method can be more robust when the depressions have obvious changes.

Abstract: Purpose The purpose of this paper is to propose a novel target automatic recognition method for unmanned aerial vehicle (UAV), which is based on backpropagation – artificial neural network (BP-ANN) algorithm, with the objective of optimizing the structure of backpropagation network, to increase the efficiency and decrease the recognition time. A hardware-in-the-loop system for UAV target automatic recognition is also developed. Design/methodology/approach The hybrid model of BP-ANN structure is established for aircraft automatic target recognition. This proposed method identifies controller parameters and reduces the computational complexity. Approaching speed of the network is faster and recognition accuracy is higher. This kind of network combines or better fuses the advantages of backpropagation artificial neural algorithm and Hu moment. with advantages of two networks and improves the speed and accuracy of identification. Finally, a hardware-in-the-loop system for UAV target automatic recognition is also developed. Findings The double hidden level backpropagation artificial neural can easily increase the speed of recognition process and get a good performance for recognition accuracy. Research limitations/implications The proposed backpropagation artificial neural algorithm can be ANN easily applied to practice and can help the design of the aircraft automatic target recognition system. The standard backpropagation algorithm has some obvious drawbacks, namely, converging slowly and falling into the local minimum point easily. In this paper, an improved algorithm based on the standard backpropagation algorithm is constructed to make the aircraft target recognition more practicable. Originality/value A double hidden levels backpropagation artificial neural algorithm is presented for automatic target recognition system of UAV.