A multi-scale temporal model for multi-view radar high-resolution range profile recognition

Radar automatic target recognition (RATR) plays a significant role in military applications. The high resolution range profiles (HRRP) contain a large amount of information, such as the distribution of the scatterers, target size and structure, so the HRRP has great prospects in the field of RATR. The traditional HRRP based RATR systems need to extract features manually, which is very difficult and complex to obtain excellent features. To solve this problem, this paper proposed a bidirectional simple recurrent unit network based on soft attention mechanism (SAMBi-SRU) to achieve HRRP target recognition. The simulation results demonstrate that the proposed model can extract robust features from the HRRP data effectively and obtain good performance and noise immunity in HRRP based target recognition.

In the radar automatic target recognition (RATR) system using high resolution range profile (HRRP), if a test sample has not been trained in training phase, it would lead to a full miss classification in test phase. In this paper, we design a classifier based on generalized confidence, which can efficiently refuse-recognize a new target. Firstly, principal component analysis (PCA) method is used to extract feature vector from every sample. Secondly, the classifier calculates each feature vector's generalized confidence on mahalanobis distance. Consequently, the distribution of confidence is used to define a refuse-recognition threshold for each training target. In test phase, for each trained-target in the database, we suppose that the test sample belongs to current target, calculate the generalized confidence, judge whether the test sample really belongs to the target or not via comparing the confidence with the target's refuse-recognition threshold. The final class is determined by vote. The experimental results demonstrate the effectiveness of the proposed algorithms.

Radar automatic target recognition (RATR) based on high resolution range profiles (HRRPs) has received intensive attention in recent years. The data-driven HRRP recognition methods based on neural networks have achieved outstanding performance with a complete target-aspect template library. However, such methods can barely achieve satisfactory performance under the condition of incomplete target-aspects or low signal-to-noise ratio. In order to solve this problem, a prior-knowledge guided neural network (PriorK-NN) is proposed for HRRP target recognition, which incorporates prior-knowledge about scattering centers and target-aspects into neural networks. Firstly, by combining the physical generative mechanism of HRRP with neural networks, we proposed an interpretable scattering center layer (SC-layer). The SC-layer can effectively reduce the adverse effects of noise and extract the location and intensity of target dominant scattering centers. In addition, we proposed a loss function named targetaspect supervised contrastive loss based on the prior-knowledge about target-aspects. By using the proposed loss function, clusters of HRRPs belonging to the same class but with different target-aspects are pulled together, while simultaneously pushing apart clusters of HRRPs from different classes. Therefore, the proposed loss function can improve the recognition performance under the condition of incomplete target-aspects. Experiments on the aircraft electromagnetic simulation dataset and the measured dataset validated the effectiveness of our proposed method on noise-corrupted HRRP recognition and demonstrated superior performance compared with other HRRP recognition methods under the condition of incomplete target-aspects.

One-dimension hierarchical local receptive fields based extreme learning machine for radar target HRRP recognition

The multilayered feed-forward neural network was applied to automatic target recognition using the high range resolution (HRR) profiles in this paper. To extract effective features from the HRR profiles, the product spectrum originally proposed for the speech signal processing was introduced to the radar target recognition community. The product spectrum was defined as the product of the power spectrum and the group delay function, which could combine the information contained in the magnitude spectrum and phase spectrum of the HRR profiles and carry more details about the shape of the aircrafts. A multilayered feed-forward neural network was selected as classifier. The HRR profiles were obtained using the two-dimensional backscatters distribution data of four different scaled aircraft models. Simulations were presented to evaluate the classification performance with the product spectrum based features. The results demonstrate that the product spectrum based features outperform the original HRR profiles and the multilayered feed-forward neural network is effective for the application of automatic target recognition of aircrafts.

A Winner- Take-All convolutional autoencoder is applied to improve the performance on the half space radar high resolution range profile(HRRP) target recognition. Feature extraction is significantly important to the radar target recognition based on the HRRP. Conventional deep models of representation learning used for HRRP target recognition commonly use the vanilla autoen-coder and deep belief net (DBN), moreover, the simulated HRRP samples used in these related work are mostly under the free space condition which is different from the real world situation. In this paper, convolution architecture autoencoder, which is more efficient in spatial feature extraction and sparse coding, is proposed. Furthermore, the half space HRRP samples, which is much more close to the real world situation and is quite different from the free space HRRP samples, is used as the dataset. Half space simulated HRRP data is used to apply the convolutional architecture on the ground target recognition and got an accuracy promotion about 7% compared to conventional vector-based module.

DARPA/Air Force Research Laboratory Moving and Stationary Target Acquisition and Recognition (MSTAR) program is developing state-of-the-art model based vision approach to Synthetic Aperture Radar (SAR) Automatic Target Recognition (ATR). The model-based approach requires using off-line developed target models in an on-line hypothesize-and-test manner to compare predicted target signatures with image data and output target reports. Central to this model-based ATR is the PEMS (Predict-Extract-Match-Search) subsystem. The Search module is critical to PEMS by providing intelligent control to traverse the hypothesis feature space. A major MSTAR goal is to demonstrate robust ATR for variations in targets including partially hidden targets. This paper will provide an update on the technology being developed under MSTAR and the status of this model based ATR research, specifically concentrating on the Search Module.

A Bayesian dynamic model based on multitask learning (MTL) is developed for radar automatic target recognition (RATR) using high-resolution range profile (HRRP). The aspect-dependent HRRP sequence is modeled using a truncated stick-breaking hidden Markov model (TSB-HMM) with time-evolving transition probabilities, in which the spatial structure across range cells is described by the hidden Markov structure and the temporal dependence between HRRP samples is described by the time evolution of the transition probabilities. This framework imposes the belief that temporally proximate HRRPs are more likely to be drawn from similar HMMs, while also allowing for possible distant repetition or “innovation”. In addition, as formulated the stick-breaking prior and MTL mechanism are employed to infer the number of hidden states in an HMM and learn the target-dependent states collectively for all targets. The form of the proposed hierarchical model allows efficient variational Bayesian (VB) inference, of interest for large-scale problems. To validate the formulation, example results are presented for an illustrative synthesized dataset and our main application-RATR, for which we consider the measured HRRP data. For the latter, we also make comparisons to the model with the independent state-transition statistics and some other existing statistical models for radar HRRP data.

High resolution range profile (HRRP) has the advantages of low complexity in imaging and processing. These features make HRRP widely adopted in radar automatic target recognition technology. Template matching is a representative technique for radar target recognition using HRRP. To overcome the attitude sensitivity problem of HRRP data, it is necessary to subdivide the training database according to attitude domain, and construct corresponding templates for matching. However, templates from traditional methods usually cause the loss of details due to the smoothing process, with limited classification performance. This paper is motivated by proposing a template generation approach based on maximum information profile, which retains the detailed information in templates for accuracy optimization. The proposed method is verified by the recognition experiments of three types of aircraft targets. The experimental results indicate that the maximum information profile outperform the mean range profile contributed by the angle domain template.

Combination of global and local filters for robust SAR target recognition under various extended operating conditions

Stable and reliable feature extraction is crucial for radar high-resolution range profile (HRRP) target recognition. Owing to the complex structure of HRRP data, existing feature extraction methods fail to achieve satisfactory performance. This study proposes a new deep learning model named convolutional neural network–bidirectional encoder representations from transformers (CNN-BERT), using the spatio–temporal structure embedded in HRRP for target recognition. The convolutional token embedding module characterizes the local spatial structure of the target and generates the sequence features by token embedding. The BERT module captures the long-term temporal dependence among range cells within HRRP through the multi-head self-attention mechanism. Furthermore, a novel cost function that simultaneously considers the recognition and rejection ability is designed. Extensive experiments on measured HRRP data reveal the superior performance of the proposed model.

The existing synthetic aperture radar (SAR) automatic target recognition (ATR) methods have shown impressive results in static scenarios, yet the performance drops sharply as new categories of targets are continuously increased. In response to this problem, this letter proposes a novel ATR method named multi-level adaptive knowledge distillation network (MLAKDN) to achieve incremental SAR target recognition. To be specific, an adaptive weighted distillation strategy is first proposed, which can alleviate the model from forgetting the knowledge of old categories by distilling multi-stage soft label information of old categories at the classification level. Then, a feature distillation method based on gradient maximum criterion is developed to filter and distill discriminative features, so as to further recall more knowledge of old categories at the feature level. Meanwhile, a model rebalancing technique is designed to effectively strike the balance of the model on new categories and old categories. Finally, a weighted incremental classification loss is presented to train the whole model. Experiments on the moving and stationary target acquisition and recognition (MSTAR) dataset and the synthetic and measured paired labeled experiment (SAMPLE) dataset illustrate that the proposed method is superior to some state-of-the-arts for incremental SAR target recognition tasks.

In the field of synthetic aperture radar (SAR) automatic target recognition (ATR), deep learning-driven methods perform well when numerous training samples are available. However, in practical SAR application scenarios, the number of available training samples is often scarce due to the limitation of observation conditions, which leads to major challenges for deep learning-driven methods in real-world deployment. Currently, the problem of ATR with only a few training samples has attracted much attention in the SAR field. In this article, we put forward an ATR method called azimuth-guided feature embedding network with dual inference mechanism (AGFEN-DIM) to achieve fewshot SAR target recognition. To be specific, the feature extraction model, i.e. azimuth-guided feature embedding network (AGFEN) is composed of the azimuth embedding module (AEM) and the dynamic feature embedding module (DFEN). Among them, AEM resorts to a set of azimuth-parameterized convolution kernels, aiming to make full use of information relevant to the current radar imaging environment. The proposed DFEN is capable of dynamically extracting the discriminative features of the target according to the input image under the guidance of azimuth-parameterized convolutional kernels. The dual inference mechanism (DIM) effectively integrates the inference results of improved prototype classification and label propagation strategy to boost the reliability and robustness of the recognition model for few-shot SAR target recognition tasks. Extensive experiments on the moving and stationary target acquisition and recognition (MSTAR) dataset demonstrate that the proposed AGFEN-DIM surpasses many state-of-the-art SAR ATR methods in various application scenarios.

Feature representation based on the high resolution range profile (HRRP) is important in radar automatic target recognition(RATR). Traditional algorithms of feature extraction utilize hallow architectures and rarely address the challenges of high-noise and unknown-noise distribution. The capability of RATA is restricted by these challenges. In this paper, a novel blind-denoising network(BDNet) is proposed to implement denoising and automatically extract features. As an extension of deep autoencoder, BDNet is based on fully convolutional architecture and employs fusion layers to transfer input features to high dimensional space. Trained with noise-to-noise, BDNet can implement blind-denoising and doesn't rely on noise distribution. Then the output of BDNet is used to classify the targets. In the experiment, we use the measured HRRP signals of four aircrafts to show the effectiveness of our methods. The results prove that BDNet can achieve blind-denoising in highnoise environment and significantly improve the performance of recognition.Andourproposed BDNetAlexNet outperforms other recognition methods.

In radar High Resolution Range Profile (HRRP) recognition field, an effective feature extraction process is the main point to improve whole recognition performance. As the targets' HRRP is the amplitude of coherent summations of complex return from scatters in each range cell, some data-driven compression methods, like PCA, LDA, etc. are widely used and achieve good results under experimental conditions, but still doesn't contribute much to solving the inherent problem of HRRP, such as small-sample problem. For these reasons, a series of SOM based feature extraction methods are introduced to explore the possibility in HRRP recognition. From machine learning perspective, these SOM based methods might more comprehensively describe the inherent structure relations within the original HRRP data distribution with its competitive learning process. In this paper, three category simulation experiments are presented to analysis the feature extraction, data storage, cluster result and classifier design. Of these experiments, SOM combined with multi-layer neural network can reduce the over fitting problem of directly apply the multi-layer neural network for HRRP recognition, which further prove the feasibility of SOM application in HRRP recognition.