This paper presents a novel mmWave-based aerial text input method that allows users to write freely through hand movements in the air, effectively recognizing English words. Unlike existing radar character recognition methods that require pauses between adjacent letters for segmentation, the proposedmmScribesystem supports continuous handwriting and streaming recognition. Our approach first utilizes Frequency-Modulated Continuous Wave (FMCW) mmWave radar to extract micro-Doppler time series features characterizing gesture movements. Then, a hybrid CTC/attention architecture baseline method is proposed to achieve end-to-end (E2E) text output without explicit letter segmentation. By introducing a dynamic block training strategy, the model can process contextual data of arbitrary length. During inference, the connectionist temporal classification (CTC) decoder generates n-best hypotheses at the character level in a streaming manner, while the attention decoder subsequently re-scores these hypotheses to update the final result at the text level. We have constructed the first word-level handwriting dataset for mmWave radar, and also conducted a comprehensive evaluation of the proposed method on this dataset. The results demonstrate that the method exhibits excellent performance. Furthermore, when deployed on mobile devices, mmScribe maintains its outstanding performance with minimal latency. Our code and dataset are available at:https://github.com/Tkwer/mmScribe.

In the scenario of ultrashort range (USR) gesture interaction, the joint range-angle parameter estimation encounters numerous challenges when using multiple-input-multiple-output (MIMO) frequency-modulated continuous-wave (FMCW) millimeter-wave (mmWave) radar. In signal modeling, the traditional far-field plane-wave assumption breaks down due to the hand motion spanning the near-field and far-field regions. Moreover, the hand target exhibits a significant angular extent in USR, making a distributed source model more realistic than a point target model. For joint parameter estimation, while the wideband FMCW signal improves range resolution, it may introduce direction-of-arrival (DOA) estimation bias and ambiguities. To address these issues, this article establishes a spherical wavefront distributed source signal model applicable to hybrid-field scenarios, which accurately reflects real-world signal propagation in USR. We analyze the errors induced by traditional signal processing algorithms, and proposes a joint range-DOA estimation algorithm for spatial localization. This method initially performs meticulous peak alignment within the range domain to correct for phase residuals from wideband signals. It then iteratively processes each range-bin data using a 2-D Capon angle estimation algorithm, which is based on the spherical wavefront MIMO radar array model, for joint estimation. Ultimately, the target’s Cartesian coordinates are computed utilizing an elliptic geometric model. Our simulations and real-data results both validate the effectiveness of the proposed model and method.

Incremental learning methods can overcome the problem of catastrophic forgetting in synthetic aperture radar (SAR) target detection models when continuously learning new class data. However, due to the characteristics of SAR images, such as large background variations and relatively stable scale differences between different target classes, existing advanced incremental detection methods perform poorly in the SAR domain. Specifically, current incremental detection research lacks adaptability to spatial distribution changes and insufficiently focuses on the model’s localization knowledge, making it significantly limited when addressing the aforementioned SAR characteristics. To tackle these issues, we propose a class-incremental learning method for SAR target detection based on smooth distribution and depth-focused knowledge distillation (SDDFD). First, we design a spatial dimension-based smooth distribution distillation (SDDL) method, which evaluates the effectiveness of spatial response points through concentration assessment and dynamically assigns weights to adapt to different background spatial distributions. Then, based on SDDL, we propose an embeddable depth-focused distillation (DFDL) approach. This approach innovatively enhances shallow localization knowledge and deep classification knowledge from a depth perspective, significantly improving the model’s localization ability. Experimental results show that our method outperforms advanced incremental detection methods in various incremental task scenarios, achieving optimal performance. For example, in one-step incremental tasks, the AP0.5 of SDDFD compared to the state-of-the-art method ERD, improved by 8.8% and 4.0% on the MSAR-1.0 and SAR-AIRcraft-1.0 datasets, respectively.

Existing incremental target detection (ITD) methods heavily depend on the diversity of information. When the scene in the new image diverges significantly from the previous training data, the detector’s ability to detect known targets diminishes considerably. The scene information in synthetic aperture radar (SAR) images is closely linked to target types, as targets of the same class often emerge in similar environments. Consequently, various scenes are frequently introduced in conjunction with new classes, posing substantial challenges to the robustness of the SAR incremental target detector. To tackle this issue, this article proposes the context-aware attention (CAA) and the Gaussian-box similarity metric (GBSM) methods to enhance the scene adaptability of SAR incremental target detectors. First, the CAA operates during the feature knowledge transfer stage, which consists of a global relation module and a local attention (LA) module. It integrates the relationship between the target and its contextual information while preserving contextual awareness through knowledge transformation. Second, the GBSM establishes a constraint factor through both 2-D Gaussian modeling and distribution similarity measurement. It further modifies the incremental localization loss to reduce the impact of target-background contrast on the model’s localization capability. We set up multiple data increment scenarios using the MSAR and SAR-Aircraft datasets. Comparative experimental results show that our method achieves better performance. In addition, the time consumption of training was recorded, and comparisons demonstrate that our method also offers advantages in efficiency.

In complex and dynamic synthetic aperture radar (SAR) scenes, few-shot detection of novel classes suffers from sample scarcity and significant distribution differences between base and novel class features, leading to severe bias and poor generalization in existing few-shot object detection (FSOD) models. To address this issue, we propose a meta-transfer learning method based on dynamic semantic guidance (DSG). This approach combines the strengths of meta-learning and transfer learning, comprising three modules: semantic guidance (SG), distribution alignment metric (DAM), and global feature dynamic aggregation (GFDA). The SG module generates guided features with query semantic information to reduce the distribution gap between base and novel classes, dynamically adapting to few-shot novel class SAR targets. The DAM module applies adversarial training to achieve dynamic feature distribution alignment, improving model bias and generalization. The GFDA module dynamically aggregates and retains critical feature information, enhancing model detection performance. Experimental results on the SRSDD-v1.0, MSAR-1.0, and SAR-AIRcraft-1.0 datasets show that the DSG method outperforms state-of-the-art methods in the SAR field (GMFBA) and the optical domain (G-FSOD), with average detection performance improvements of 1.21%, 1.45%, 1.44%, and 9.76%, 2.86%, 1.8%, respectively.

The benefits demonstrated in smart home and energy-saving applications have sparked considerable interest in non-contact technologies for detecting indoor human presence. In opti-mizing energy utilization, precise remote assessment of individual count is of paramount importance.However, the presence of multiple human targets in indoor environments poses a challenge for quantity estimation, especially when using a simple one-transmitter-two-receiver 24GHz radar system. Due to limitations in bandwidth and the number of receiving antennas, the system exhibits poor resolution, rendering traditional distance-angle estimation algorithms ineffective in distinguishing targets.To address the indistinguishability issue of multiple targets in the distance-angle domain, we propose a target counting algorithm based on the micro-Doppler domain. Leveraging an atomic norm minimization method, this approach separates target respiratory frequencies and determines the number of targets. Through simulations and practical experiments, our approach demonstrates superior performance in target counting compared to traditional Fourier transform and compressive sensing methods.

In this paper, we present a model integrating joint sparsity and sinusoidal fitting to address the challenge of static human detection in indoor environments with interference. The motivation behind this model lies in leveraging the frequency domain sparsity induced by the micro-Doppler effect caused by human breathing. The tasks of interference suppression and human detection are formulated as an optimization problem, fitting the signals using sinusoidal functions with sparse coefficients. To effectively address this challenge, we introduce an iterative algorithm based on Improved Sparse Matching Pursuit (ISMP) within the compressed sensing framework, aiming to capture the micro-Doppler information of the target, fit the low-dimensional structure of human respiration in the phase domain, and simultaneously suppress interference micro-motion features. Compared to conventional methods using max average power and standard deviation , our model effectively suppresses interference and enhances target signals in real single-channel radar experiments, even amidst mixed interference with human targets.

Millimeter wave radar can be used for monitoring vital signs of personnel under non-contact conditions, and has stronger anti-interference and privacy capabilities. This article proposes a method based on the 60GHz FMCW millimeter wave radar system, which utilizes an improved ResNet50 network to classify six respiratory patterns of the human body (Eupnea, Tachypnea, Bradypnea, Biots, Cheyne Stokes, and Central Apnea) based on the echo waveform. This method demonstrates high accuracy and robustness in respiratory classification in complex environments; In addition, we also use the network reasoning optimization method to deploy it on edge computing devices; According to the experiment, when the refresh rate of edge computing system is 20Hz, the recognition accuracy of this method is up to 95.25%, which verifies the accuracy and robustness of this method.

Traditional supervised learning methods have achieved great success in automatic target recognition (ATR). Unfortunately, if the model is trained exclusively on new class samples, the model will forget all knowledge about the old class samples. This phenomenon is called catastrophic forgetting. Incremental learning method can prevent catastrophic forgetting by keeping a small number of old class samples as exemplars and training them together with new class samples. However, the ratio of old to new class samples is still seriously out of balance. In this paper, an data augmentation method of old class samples is proposed by using random interpolation (RI) between two exemplars to generate fake samples in the process of incremental learning. In this method, the distribution of the original classes is partially restored while also creating a numerical balance between the old and new class samples. Experiments on the Moving and Stationary Target Acquisition and Recognition (MSTAR) dataset demonstrate the effectiveness of this method in Incremental SAR ATR.

In the task of class-incremental learning in SAR object detection, traditional deep learning training methods exhibit numerous limitations. However, updating the original detection model with only new data leads to the problem of catastrophic forgetting. To address these challenges, this paper proposes an incremental detection method based on adaptive distributed response distillation. This method effectively preserves the model’s localization and classification knowledge for old classes through the utilization of a response discreteness-based selection strategy and smooth mask-weighting mechanism. Experimental results on the MSAR dataset demonstrate that our approach achieves a 2.7% increase in average precision(AP) compared to the state-of-the-art method ERD. The proposed method can effectively mitigate catastrophic forgetting and provide higher detection accuracy.

With the continuous development of deep learning, a large number of deep learning models are gradually emerging in the field of SAR target recognition. However, SAR targets usually have complex textures and noises, and it is difficult to directly extract effective feature information and rich contextual information between the target object and the background, which inhibits the potential of deep learning models to further improve the recognition ability in the SAR domain. To ameliorate this problem, a GIC-Mechanism that improves the ability to capture global and local feature information interactively is proposed in this paper and applied to the Resnet family of models. In the recognition task of two SAR target datasets, the mechanism designed in this paper with the ability of multi-scale losing feature information interaction and cross-feature mapping layer information interaction improves the recognition performance of the Resnet series model by 2.16%-2.81%, which is effective.

Normal human heartbeat and respiratory signals are not steady sinusoidal signals and contain a multitude of harmonic components. The principle of vital sign detection using millimeter-wave radar is based on monitoring the chest cavity’s movements to obtain vital sign signals. Due to the amplitude of the respiratory signal being an order of magnitude higher than that of the heartbeat signal, the higher-order harmonics of the respiratory signal, particularly the second and third harmonics, which fall within the frequency range of the heartbeat signal, can easily overshadow it. This overlap creates difficulty in separating respiratory and cardiac signals using traditional methods. This paper proposes a heartbeat signal extraction method based on template selection, which extracts the heartbeat signal during respiratory pauses, suppresses the interference of respiratory harmonics, and experimentally validates the effectiveness of the method.

Synthetic aperture radar (SAR) imagery is susceptible to multiple factors, including radar parameters, imaging mode, and imaging angle. Consequently, the resolution of the training and test data often differs, leading to the failure of existing deep learning-based SAR target recognition methods. To address this challenge, this paper introduces a transformer-based cross-resolution target recognition method for SAR images based on feature alignment and aggregation. The objective is to improve the accuracy of cross-resolution target recognition. Experimental results obtained from OpenSARShip, an openly available dataset with ship images captured at various resolutions, demonstrate the superior performance of the proposed method compared to the current state-of-the-art (SOTA) technique.

In recent years, the rapid development of synthetic aperture radar (SAR) technology has enabled researchers to obtain SAR targets from multiple sources within the same class. However, most existing SAR target recognition methods are designed for single-source target recognition. When applied to multi-source target recognition in SAR imagery, the different data distributions from various sources can negatively impact recognition accuracy. To address this issue, this paper proposes a scale distillation method for multi-source target recognition in SAR imagery. This approach involves extracting feature representations of the target, incorporating multi-scale semantic information through the multi-scale feature fusion mechanism (MSFFM), and performing knowledge distillation on the output scale features to enhance discriminative knowledge about the target scales. Experimental results on the FUSAR-Ship and OpenSARShip datasets from different sources demonstrate an average improvement in classification accuracy by 2.5% after applying scale distillation, which is a significant advantage over other methods.

Millimeter-wave radar has found widespread applications in perception domains such as gesture recognition. In comparison to single-hand gestures, radar echo signals from multi-hand gestures involve the superposition of multiple signals, presenting a more challenging recognition task. This paper introduces a multi-hand gesture signal separation method based on the enhanced Multiple Signal Classification (MUSIC) algorithm, capable of simultaneously distinguishing the motion states of the left and right hands. Specifically, this paper utilizes Minimum Variance Distortionless Response (MVDR) to extract joint angle-range information for gesture features. The enhanced MUSIC algorithm is then applied to reconstruct single-hand gesture Range-Angle Map (RAM) by constructing a new noise subspace. The effectiveness of the proposed algorithm is validated through simulations and radar signal experiments, demonstrating its capability to separate multi-hand gesture signals from the foundation of multi-hand gesture RAM to obtain single hand RAM.

Deep learning methods have been widely used in the field of SAR target detection, but conventional deep neural networks do not have the ability to preserve old knowledge. With new classes of SAR images constantly being acquired, typically we re-train the network using all the data, which puts a burden on storage and computing. However, if the model updates using only new data, the detector will catastrophically forget previous knowledge. To address this problem, this paper proposes a Class-incremental learning (CIL) method for SAR multi-class target detection, based on an overall framework of knowledge distillation (KD), which enables to maintain the ability of the detector to detect old classes of targets whilst learning new knowledge. The method includes two core components: First, we introduce the Localization Distillation (LD) into the incremental learning framework, preventing the impact of multi-scale SAR target introduction on small target detection capability during the incremental learning process; Second, we apply the global relationship module (GR) to the feature-review process, which strengthens the learning of target-background correlation. We performed comparison experiments on the MSAR-1.0 SAR multi-class target detection dataset, and our method has better performance compared with several mainstream incremental detection methods.

Radar specific emitter identification aims to recognize individual radar emitters based on subtle differences in transmitted signals. This paper proposes using differential constellation figure (DCF) features extracted from received radar signals for this task. A compact deep neural network called AL-ResNet classifies the DCF features to identify individual radar emitters. Experiments on simulated radar signals show the proposed DCF method achieving 91.76 % accuracy in recognizing 6 radar emitters using ResNet50 architecture, outperforming techniques using other signal features. Nevertheless, experiments on the lightweight model ALResNet shows efficient deployment of the utilized network.

The extensive adoption of deep learning techniques in SAR image interpretation has spurred a heightened interest in the explainability of these models. SHAP, a masking-based explainability method, relies on conventional occlusion techniques (baseline values) that are frequently overly rigid and introduce varying degrees of interference. To faithfully capture the influence of distinct regions in SAR images on the overall interpretation, we introduce a baseline value algorithm based on multivariate interactions and SAR clutter distribution. This algorithm is applied to SHAP attribution. Experimental results on the MSTAR dataset illustrate that the baseline values computed by our algorithm effectively eliminate all existing significant patterns in the original samples without introducing new patterns.

Deep Learning (DL) models have proven effective in Synthetic Aperture Radar (SAR) images classification. However, the black-box nature of deep learning models hinders explainability. A common approach to compute the attributes of input variables in a sample involves masking some input variables of a deep neural network (DNN) and measuring the output variation of the masked input sample. Typically, baseline values of input variables are used to mask the input variables. Recent research suggests that the effectiveness of masking methods with different baseline values. In this paper, we compare Zero, Mean, and Random baseline values based on the SHAP method and provide guidance for practitioners in selecting an appropriate masking method for similar tasks.

This paper presents an innovative approach for detecting human vital signs in multi-target scenarios using Frequency-Modulated Continuous-Wave (FMCW) radar within a Multiple-Input Multiple-Output (MIMO) framework. Recognizing the limitations of traditional contact-based vital signs monitoring methods, especially for sensitive groups such as infants or burn victims, our study proposes a non-invasive, radar-based solution. We develop a Linear Mixed Model (LMM) to address the challenges posed by multi-target, multi-path aliasing interference, employing Blind Source Separation (BSS) model and FastICA algorithms for signal separation and extraction. The algorithm’s effectiveness is validated through simulation experiments, where we successfully distinguish and analyze vital signs, such as heart and respiration rates, from phase-modulated Intermediate Frequency (IF) signals. The IWR6843AOP MIMO radar, operating in the 60-64GHz band, is utilized for data collection. Our results demonstrate the potential of this methodology in accurately identifying vital signs in complex, multi-target environments. Furthermore, the study suggests enhancements to the existing signal processing flow, highlighting areas for future research in real-time vital signs monitoring using FMCW radar.

This paper presents a comprehensive study on the application of Frequency-Modulated Continuous-Wave (FMCW) radar in vital sign monitoring. We identify and address the limitations of traditional signal-to-noise ratio (SNR) metrics in human vital signs detection field. Recognizing the challenges posed by the human body’s orientation, distance from the radar, and random body movements, we propose an innovative SNR estimation method. Our approach considers various real-world scenarios, including different target distances, angles, and levels of body movement, to ensure the robustness and effectiveness of vital sign detection. Through a series of experiments, we demonstrate the efficacy of our novel SNR algorithm, underscoring its superiority in different situational contexts compared to conventional methods.

As an essential task in remote sensing field, the result of road extraction on Synthetic Aperture Radar (SAR) image is still unsatisfactory, especially for the low resolution SAR images. A potential reason is the extremely low ratio of road pixels to the background ones, obstructing the network from extracting sufficient information for road extraction. To address this issue, we propose an approach to combine dense connection and U-Net to fully exploit the potential information of road regions. In addition, attention mechanism is introduced to guide the network focus more on the region-of-interest. Comprehensive experiments on our own dataset demonstrate that the proposed approach outperforms state-of-the-art methods by large margins on the road extraction task, especially for the low resolution images.

Synthetic aperture radar (SAR) targets are often characterised by high manoeuvrability and strong concealment, resulting in scarce and few-shot SAR data. Due to the scarcity and variability of samples causing significant fluctuations in class centers, the sample distribution is challenging to determine, resulting in the model’s inability to accurately represent the potential representative features of few-shot SAR targets. Therefore, we propose a few-shot SAR target detection method based on Gaussian meta-feature directed aggregation (GMDA) using the meta-learning paradigm. Specifically, a Gaussian distribution is constructed on the support branch to estimate the class distribution of the few-shot SAR target and replace the traditional class prototype. Based on this, we propose the feature information maximization module (FIMM) to avoid the bias of the feature information and achieve the directional expression of the features in order to complete the efficient aggregation. Experiments on SRSDD-v1.0 and MSAR-1.0 datasets show that our method almost beats the state-of-the-art methods at this stage and achieves state-of-the-art performance in all settings.

In-air handwriting necessitates consistent motion tracking, in contrast to millimeter-wave (mmWave) radar-based simple gesture recognition techniques. However, during long-duration gesture tracking, challenges, such as body motion interference and environmental clutter, become more pressing. Moreover, due to the lack of a supporting surface in in-air handwriting, slight arm tremors also can result in unsmooth trajectories. To address these challenges, this article proposes a two-stage processing framework called MMHTSR. In the first stage, the state-space equations are reestablished, and a locally correlated 2-D Gaussian process regression (GPR) algorithm is employed for interframe prediction. By incorporating uncertainty estimation, weights are assigned to the next frame data, effectively suppressing interference from nongestural targets. In the second stage, real-time smoothing and tracking of gesture trajectories are accomplished using a Kalman filter, followed by mapping the trajectories onto the Cartesian coordinate system. Finally, an end-to-end signal processing framework is deployed on a low-cost 60-GHz mmWave radar prototype, and gesture trajectory recognition is achieved using deep learning methods. Experimental results demonstrate that MMHTSR can accurately track motion gestures within the range of approximately 5–40 cm and successfully recognize 30 classes of in-air gesture trajectories, including uppercase letters A–Z and four interactive gesture actions. Furthermore, the proposed framework exhibits robust performance across various scenarios which shows its adaptability.

To enable the synthetic aperture radar (SAR) automatic target recognition (ATR) system to continuously adapt to new recognition scenarios, it is necessary to equip the system with the ability to quickly update models. However, when these models learn new tasks, the knowledge of old tasks is quickly forgotten, a phenomenon known as catastrophic forgetting. The reason for catastrophic forgetting is that the model does not use the features of old tasks sufficiently. In this letter, an exemplar-free class incremental learning based on maximizing mutual information (CIL-MMI) is proposed to solve this problem. To effectively use the extracted features, CIL-MMI actively clusters features to maximize the mutual information (MI) between features and corresponding labels. The proposed method successfully avoids the distribution overlap caused by the small interclass differences and large intraclass variances inherent in SAR images. Experiments on the Moving and Stationary Target Acquisition and Recognition (MSTAR) dataset indicate that the proposed method outperforms state-of-the-art approaches, demonstrating improvements of 5.41%, 1.93%, and 2.47% at incremental steps 1, 2, and 3, respectively.

The features employed for synthetic aperture radar (SAR) target recognition have evolved from traditional SAR target geometric features and pattern features to modern deep features, indicating a trend of increasing recognition accuracy but decreasing feature interpretability. Therefore, the fusion of multidimensional features has been investigated by many researchers. Existing feature fusion methods typically involve simple concatenation or addition of geometric features and pattern features with deep features, or directly incorporating them into deep networks. However, such fusion methods mentioned above inadequately consider the potential conflicts between features and hard to fully exploit multidimensional features. To solve the above problem, a multidimensional feature joint learning framework (MFJL-Framework) that serves the SAR target recognition task is proposed in this article, which consists of three models. Specifically, the SGC-GA-Model can select pattern features for SAR targets based on geometric feature constraints, the Global and local Feature Information interaction Capture model (GFIC-Model) can select deep features with high-level abstract semantics, and the MFFS-Model can complement and fuse these two types of features to maximize the utilization of feature information. Experiments and comprehensive ablation studies on four datasets, namely OpenSARShip-1.0, FUSAR-Ship, MSTAR-T72Variants, and SAR-AIRcraft-1.0, collectively demonstrate that the recognition performance of our proposed FJL-Framework outperforms the current state-of-the-art methods.

Synthetic aperture radar (SAR) target recognition plays an indispensable role in interpreting SAR images. However, differences in radar parameters (including factors such as imaging modes and imaging angles) often lead to resolution differences between training and test data, posing challenges for existing methods in recognizing SAR targets under cross-resolution conditions. To address this issue, this article proposes a domain adaptation (DA) framework based on feature aggregation and compensation (FAC) for cross-resolution target recognition in SAR imagery. Initially, we employ a unique local vision transformer (LocalViT) to establish global and local adversarial networks that capture invariant features under cross-resolution conditions. Following this, we design a multiscale feature fusion module (MSFFM) to capture multiscale semantic features of targets at different resolutions. Subsequently, we propose a novel class feature aggregation module (CFAM) to map targets of varying resolutions onto the unit sphere, thereby aggregating features of samples from the same class and distinguishing those of samples from different classes. Finally, we narrow down the difference in frequency-domain information of targets at different resolutions by developing a resolution semantic compensation module (RSCM). This module compensates for the semantic feature about resolution during target recognition across varying resolutions by converting high- and low-frequency information. The experimental results on three SAR datasets (OpenSARShip, FUSAR-Ship, and SRSDD-v1.0) confirm that our method outperforms the state-of-the-art (SOTA) DA methods, with an increase of 2.06%, 1.96%, and 1.61% in three sets of cross-resolution scenes, respectively.

Deep neural networks have shown remarkable effectiveness in SAR target recognition. However, the explainability problem for deep neural networks remains insufficiently addressed. One approach to tackle this challenge is the SHAP method. It enhances the explainability of deep neural networks in SAR target recognition by observing how the target, shadow, and clutter regions play their own distinct roles. The masked regions are typically filled with Zero, Mean, or Random values in optical images. But if the same operation performed on SAR images, it will affect the distribution of clutter and thus introducing new out-of-distribution challenge. In this paper, we propose a novel masking method to enhance the reliability and efficiency of the SHAP method in SAR-ATR applications. Experimental results on the MSTAR and OpenSARShip-1.0 datasets demonstrate that our proposed method provides a more faithful representation to show the importance of every single regions in SAR target recognition. Compared to methods using Zero values, Mean values, and Random baselines, our proposed method significantly enhances the reliability of explainability.

The gradual increase of SAR imagery often accompanies the appearance of new targets, but traditional detection frameworks can only detect existing target classes and cannot detect new. Typically, we must update the model using both new and old data, which puts a strain on storage and computation, but if we only update with new data, the detection performance on old classes will suffer dramatically. For this reason, this paper proposes to use the continual learning (CL) method to solve the problem of SAR target incremental detection. Mainstream CL methods generally consider localization to be a class-irrelevant function, however, this strategy is unsuitable for SAR imagery with significant background changes, leading to poor detection performance. Addressing the above issues, this paper proposes a continual learning object detection (CLOD) method, with an overall framework based on knowledge distillation. The focus of the methodology consists of two parts: firstly, we introduce the predicted location probability representation (PLPR) method, by using spatial discretization and segmented probabilistic statistics, to transform the localization results into probability distributions, thus allowing the localization function to participate in the continual learning process; secondly, we design a proposal selection strategy according to the background characteristics of SAR images, which improves the quality of the proposals during the knowledge review to further optimize the learning effect. Experiments on the latest multi-class SAR target detection dataset MSAR-1.0, show that our method is able to learn new knowledge with less performance penalty for old classes than other methods. In multiple data incremental settings, our method provides a 2%-11% performance improvement over numerous common methods.

Synthetic Aperture Radar (SAR) targets often exhibit characteristics such as high mobility and strong concealment, resulting in scarce SAR data and the manifestation of few-shot data properties. These few-shot SAR targets are susceptible to interference from complex background information and mutual interference of target features, making it challenging to distinguish SAR targets from the background. Additionally, there is confusion in features among different targets, leading to models being highly insensitive to few-shot SAR targets under complex distribution conditions in new tasks. Similarly, these few-shot SAR targets exhibit significant sample scarcity and sample variations, resulting in pronounced fluctuations in class centers and difficulty in determining sample distributions. This leads to challenges in accurately representing the potential representative features of few-shot SAR targets by the model. To address these issues, further enhancement of SAR target features is necessary to provide a robust foundation for the ultimate aggregation module. Therefore, based on the meta-learning paradigm, we propose a method for few-shot target detection in SAR imagery via intensive meta-feature aggregation (IMFA), aiming to reinforce SAR target features for improved representation. Specifically, firstly, we propose a novel hierarchical multi-head cross attention (HMCA) to capture global multiscale contextual information in different subspaces and analyze representative features between different targets to distinguish SAR targets from the background. Then, based on HMCA, we introduce a novel feature coupling module (FCM) to couple support features with cognitive information from the query image on the support branch. This is done to reduce the confusion and mutual interference of features between targets while enhancing the model’s generalization ability on new tasks. Finally, on the support branch with query-aware information, we construct a Gaussian distribution to estimate the class distribution of few-shot SAR targets and replace traditional class prototypes. On this basis, we propose the feature information maximization module (FIMM) to avoid feature information shift, greatly strengthening the expression of potential features. Through these steps, reinforced meta-features can be obtained, enabling efficient aggregation. Experiments on the SRSDD-v1.0 and MSAR-1.0 datasets demonstrate that our method has consistently outperformed state-of-the-art approaches in all configurations, achieving state-of-the-art performance.

Recently, people have become more concerned about life and health, and are paying more attention to the detection of their vital signs indicators. Non-contact detection of respiration rate(RR) and heart rate (HR) by using millimeter-wave (mmWave) radar is common. In this article, we propose a detection system based on multi-vital box fitting approximation using millimeter wave radar. On the basis of using different vital boxes signals with a certain period correlation, a new vital signal can be obtained by joint multi-boxes fitting approximation to a new vital signal, and taking measurements of respiration rate and heart rate. The experimental results show that the accuracy of the system using the multi-vital box vital signal fitting approximation improves about 3% for the detection of RR and about 2% for the detection of HR than the traditional single vital box system.

In recent years, some studies used computer to generate synthetic aperture radar (SAR) images. Although the synthetic SAR image looks realistic, there is a domain gap and distribution differences between synthetic SAR images and measured SAR images which makes it difficult to use synthetic SAR images directly. The image translation network can reduce the gap between different domain images, but existing optical translation networks are hard to correctly learn and translate the background information of SAR images. In this paper, a synthetic-to-measured SAR target image translation network, called SAR-UT network, based on U-NET network and Transformer is proposed to bridge the domain gap between synthetic and measured SAR target images. The proposed method can extract and fuse the texture features of each layer more effectively, so as to translate SAR images with more realistic details and higher image quality. Experiments on Synthetic and Measured Paired Labeled Experiment (SAMPLE) dataset show that the proposed translation network can well learn the potential features of SAR image and translate the synthetic SAR image into the image closer to the measured SAR image. Experimental result shows that the performance of SAR-UT is better than the state-of-the-art methods.

Vital signs monitoring technology is an important part of modern healthcare. With the continuous advancement of medical technology, the existing vital signs monitoring technology can no longer meet people’s needs. Through the analysis of different application scenarios, it can be seen that the new vital signs monitoring technology presents a trend of non-sensory monitoring, long-term monitoring, bed monitoring, and early diagnosis. Remote vital signs monitoring using millimeter wave radar has the advantages of non-contact, continuous and high degree of freedom, and can be used to monitor the vital signs of special patients. However, the signal received by mm-wave radar are very sensitive to random body movements, which reduces the accuracy of heart rate and respiratory rate. To overcome this challenge, we propose a method based on chest mechanical motion modeling to remove random body movements.

The availability of a large amount of labeled data has promoted the success of deep learning in Synthetic aperture radar (SAR) target recognition tasks, but these labeled data are invariably obtained manually, which requires huge labor costs. With the development of the SAR field, SAR target categories are still increasing, so there is a large amount of data to be marked. In this paper, we use labeled data to discover novel categories, thereby reducing the cumbersome labeling process. We innovatively propose to apply AutoMix to the labeled dataset to expand the neighborhood distribution and motivate the discrete sample space continuously, which greatly improves the generalization ability of the model. Three stage learning is used to solve the problem of novel class discovery. Firstly, self-supervised learning is used to learn common features from labeled data and unlabeled data. Secondly, AutoMix is used to labeled data which is used to train the feature extractor by supervised learning. Finally, the knowledge of labeled data is transferred to unlabeled images to generate pairs of pseudo-labels for clustering. We experimentally proved that our model can effectively discover novel categories in unlabeled data.

Synthetic aperture radar (SAR) images are difficult to acquire, and the number of images of different targets often varies greatly, resulting in a large number of imbalanced class distributions in practical applications. Existing classification models usually focus too much on the classes with a large number of samples and less on the minority classes with higher-value, which leads to the degradation of classification performance. A novel method for imbalanced classification in SAR images based on reinforcement learning adaptive cost adjustment is proposed in this paper. The method can adaptively search a cost factor and adjust it continuously according to the predicted classification effect so that the performance of all classes of samples is the best possible. This method can obtain a more accurate cost factor and modify the decision boundary of the classifier to achieve a more accurate classification of the minority classes. Experiments on the MSTAR dataset show that the proposed method achieves binary and multi-classification, significantly alleviates the effect of imbalanced data, shows better classification results on all kinds of samples, and outperforms existing methods in overall performance.

Automatic target recognition (ATR) of Synthetic Aperture Radar (SAR) images is significantly important for SAR image interpretation. However, most ATR methods are data-driven, which perform not well when the dataset is small. In this paper, considering that similarities in features among SAR images of different categories captured by the same SAR, we propose a knowledge-assisted few-shot SAR ATR method to leverages features of SAR images of other categories. Besides, a scheme that applies ACGAN to extract information from the auxiliary dataset to improve the performance of classification tasks on the target dataset is presented. Experiments on the MSTAR dataset demonstrate that integrating knowledge of auxiliary dataset can significantly boost the accuracy of SAR ATR by 5%-10%, especially when the target dataset is small.

In order to rapidly improve the automatic target recognition (ATR) system when new unknown samples are constantly captured, it is necessary to examine the existing training samples and recognition model so that the ATR system could autonomously assess new unknown samples with low predictive reliability during the recognition process and learn them preferentially. Incremental learning methods generally consider forming key exemplar set from existing known samples, but rarely managing updates of unknown samples. In this paper, an incremental samples’ evaluation and management method from the perspective of distribution reliability (DRaIL) is proposed, which realizes the retention of existent reliable exemplars and the predictive-reliability-assessment-based updating of new unknown samples simultaneously. DRaIL preserves the prior distribution in the high-density and overlap regions first, and then the classification reliability and “in-of-distribution" reliability of new unknown samples are evaluated based on the consistency between the new and the preserved distribution. Updating the new samples with low reliability using new labels could rapidly improve the classification surface and add new classes. Experimental results for the practical incremental learning scenario demonstrate the validity of the proposed DRaIL on representative exemplar selection and reliability ranking performance.

The traditional Synthetic Aperture Radar Automatic Target Recognition (SAR/ATR) algorithm can train a sufficient number of known class samples and classify the samples in the test set. However, if the old model is trained only with the new class samples, the old class samples’ knowledge is easily forgotten by the new model, which is called catastrophic forgetting. The reason is that the model only fits the distribution of current training samples, so training the whole data set is necessary. Due to the limitation of storage resources, it is often not feasible to retain the whole data set. In order to avoid this phenomenon, a small number of old class samples can be kept to train with the new class samples. Therefore, how to select the old class samples becomes the key point. In this paper, the Density Coverage-Based Exemplar Selection (DCBES) is proposed to choose the key samples of the old class. DCBES selects samples based on the metric learning theory and the set covering theory. First, the metric learning theory is used to measure the similarity between samples and to obtain the density range of samples. Then the exemplar selection problem is considered a set covering problem, to select a fixed number of exemplars to achieve the maximum coverage of the class density range. Experiments on the Moving and Stationary Target Acquisition and Recognition (MSTAR) dataset show that our method is superior to other exemplar selection methods and achieves the best results.

In recent years, deep convolutional neural networks (DCNNs) have been widely used in the task of ship target detection in synthetic aperture radar (SAR) imagery. However, the vast storage and computational cost of DCNN limits its application to spaceborne or airborne onboard devices with limited resources. In this paper, a set of lightweight detection networks for SAR ship target detection are proposed. To obtain these lightweight networks, this paper designs a network structure optimization algorithm based on the multi-objective firefly algorithm (termed NOFA). In our design, the NOFA algorithm encodes the filters of a well-performing ship target detection network into a list of probabilities, which will determine whether the lightweight network will inherit the corresponding filter structure and parameters. After that, the multi-objective firefly optimization algorithm (MFA) continuously optimizes the probability list and finally outputs a set of lightweight network encodings that can meet the different needs of the trade-off between detection network precision and size. Finally, the network pruning technology transforms the encoding that meets the task requirements into a lightweight ship target detection network. The experiments on SSDD and SDCD datasets prove that the method proposed in this paper can provide more flexible and lighter detection networks than traditional detection networks.

As a sensor, millimeter-wave (mmWave) radar can realize the function of touchless gesture control, and it has become a hot research spot in the field of Human-Computer Interaction (HCI). This paper proposes a robust mmWave gesture recog-nition method, which can recognize gestures end-to-end with high accuracy in a complex environment. It is worth mentioning that the Angle-Range joint temporal (ART) feature is extracted from radar echoes to describe gestures, which is a 3D matrix feature including azimuth, distance and speed in-formation. Then, the CNN-LSTM network is used to real-ize gesture classification. The experimental results show that this method has an accuracy of 98.5% for the recognition of four gesture types. The robust performance of the proposed method is validated by data samples collected in complex en-vironment and random population, and the average recognition accuracy remains above 88.7%.

Due to the defects of millimeter wave (MMW) radar in angle resolution, with the increase of detection distance, the ability of radar to distinguish adjacent objects in azimuth direction will be weakened, resulting in the loss and misestimation of objects information. To solve this problem, radar and optical image object detection methods are fused. The information obtained from the optical image data is used to provide a prior information for radar signal processing algorithm, so as to improve the performance of radar object detection system. Based on the simulations and experiments in a variety of multi-object moving scenes, it shows that the fusion method can improve the detection accuracy of the system and restore the spatial location of the objects more accurately than traditional ways.

Although deep learning has achieved great success in automatic target recognition, the model needs a large number of labeled samples for training. In real life, it is a time-consuming and laborious work to label unlabeled samples, so how to find unknown classes from a large number of unlabeled samples has aroused widespread concern. In this paper, we study how to discover unseen class from an unlabeled image set under the assumption that there are samples related to the unseen class but of different classes as prior knowledge. The Automatic Unseen Class Discovery (AUCD) algorithm is proposed in this paper, which mainly solves the problem of unseen class discovery from two aspects, one is how to actively form clusters according to their classes for unknown class samples, and the other is how to obtain the number of formed clusters. Several experiments based on the Moving and Stationary Target Acquisition and Recognition (MSTAR) dataset prove the effectiveness of the proposed approach in the field of unseen class discovery.

The huge computing and storage requirements of deep con-volutional neural networks (DCNNs) limit their application on edge computing devices. In this article, we propose an attention mechanism based on the feature map quality evaluation algorithm (IQE). The knowledge distillation method based on the IQE attention mechanism uses the IQE method to identify important knowledge in the pre-trained SAR target recognition deep neural network. Then in the process of knowledge distillation, the lightweight network is forced to focus on the learning of important knowledge. Through this mechanism, the method proposed in this paper can efficiently transfer the knowledge of the pre-trained SAR target recognition network to the lightweight network, which makes it is possible to deploy the SAR target recognition algorithm on the edge computing platform. Comparison experiments with several commonly used knowledge distillation methods have proved the effectiveness of our proposed method. In addition, we also verified the performance of the lightweight network obtained by our method on the edge platform based on the K210 processor.

In synthetic aperture radar (SAR) images, there are a large number of dense multi-scale objects, especially dense multi-scale ships docked along the coast. Existing object detection methods are difficult to simultaneously detect dense multi-scale objects in complex background. A novel method for dense multi-scale object detection in SAR images based on Feature-Transferable Pyramid Network (FTPN) is proposed in this paper. In the stage of feature extraction, the feature maps of each layer are connected effectively and the feature maps of various scales are extracted. This method can extract the features of dense multi-scale objects more effectively, so as to realize simultaneous detection of dense multi-scale objects in SAR images. Experiments on SSDD dataset, AIR-SARShip-2.0 dataset and Gaofen-3 dataset show that the proposed method can achieve dense multi-scale object detection, and the overall performance is better than the state-of-the-art methods.

With the maturity of synthetic aperture radar (SAR) technology, the problem of imbalanced data has gradually emerged. This problem makes it difficult for the automatic target recognition (ATR) model to properly learn the classification boundaries of majority and minority category target samples. In this article, we propose an ATR model with new architecture, called the cost-sensitive awareness-based automatic target recognition (CA-ATR) model, which provides an effective way of solving the problem of imbalanced data. Aimed at the two issues caused by imbalanced data on ATR models, the proposed method solves the problems from both the data and algorithm levels. At the data level, CA-ATR avoids adverse correlations among the target samples through different oversampling methods. By making the ATR model cost-sensitive, the proposed method also avoids the empirical risk preference of the ATR model for majority category target samples at the algorithm-level. At the same time, CA-ATR can autonomously learn different cost-sensitive awareness from different imbalanced data sets. The awareness enables the ATR model to more accurately learn the classification boundaries between target samples that belong in different categories. Several experimental results show the superiority of the proposed approach based on the Moving and Stationary Target Acquisition and Recognition (MSTAR) data set. Compared with other imbalanced learning methods, the proposed method is able to solve different types of imbalanced data problems.

Since there are differences in the natural frequency of various synthetic aperture radar (SAR) target samples in reality, the problem of imbalanced data on the automatic target recognition (ATR) model has gradually appeared in recent years. The problem makes the classification boundary learned by the ATR model often fuzzy or even wrong. In this article, an SAR target sample generation method was proposed, called demand-driven generative adversarial nets (DDGANs), which provided an effective way to implement imbalanced data learning. When the imbalanced data exacerbated the deterioration of the minority category target samples distribution, the proposed method generated samples to alleviate this negative impact. The proposed method innovatively used two convolutional neural networks to form the discriminator of DDGAN. Among them, a convolutional neural network was used to determine whether the generated sample is real or fake. Moreover, another convolutional neural network can simultaneously dig out the generation demands of different categories of target samples when recognizing the generated samples. The generation demands enabled DDGAN to allocate different generation capabilities to different target samples on demand, thereby alleviating the negative impact of data imbalance. At the same time, DDGAN can autonomously learn the generation demands from imbalanced training sets. Several experimental results based on the Moving and Stationary Target Acquisition and Recognition (MSTAR) dataset showed the advantages of DDGAN. Compared with existing imbalanced learning algorithms, the proposed method had obvious superiority in recognition performance and data generation efficiency.

Although deep learning methods have achieved great success in synthetic aperture radar automatic target recognition (SAR ATR), their accuracies decline sharply as new classes are learned, which is known as catastrophic forgetting. The overlapping or confusion between the representations of new and old classes in the feature space is the main cause of catastrophic forgetting. In this paper, the Incremental Class Anchor Clustering (ICAC) is proposed to address this issue. ICAC solves this problem from three perspectives: first, how to learn the new classes; second, how to enable the model to recognize and classify the old classes; third, how to solve the imbalance between old classes and new classes. To learn the new classes, ICAC adaptively adds new anchored class centers for new classes, and the features of each new class will be clustered around the corresponding anchored class center. To enable the model to recognize and classify the old classes, ICAC stores some exemplars for the old classes to ensure the classification ability of the old classes without losing the old class centers in the feature space. At the same time, ICAC adopts knowledge distillation to further alleviate catastrophic forgetting. To solve the imbalance between old classes and new classes, ICAC proposes a learning strategy named Separable Learning (SL), which computes the losses of the old and new exemplars separately and then adds the two losses to make a gradient descent. Experiments on the MSTAR dataset and OpenSARShip dataset demonstrate the effectiveness of this method in SAR automatic targets recognition.

The rapid development of generative adversarial nets (GANs) has led to an increasing number of applications for the synthetic aperture radar (SAR) automatic target recognition (A-TR) with a small sample set in the past few years. However, the generated samples by the GAN s sometimes even lead to a decrease in the performance of the ATR model. In this paper, we propose a filtering approach to address this harm of generated samples. The proposed filtering approach is based on a stable generation model. The stable generation model can continuously and stably generate different batches of target samples. Then, multiple SVMs trained by different SAR target sample sets provide pseudo-labels to the other SVMs to improve the accuracy of the filtering results. Therefore, the proposed approach improves the recognition ability of the A-TR model dynamically while continuously filtering generated target samples. Several experimental results show the superiority of the proposed filtering approach based on the Moving and Stationary Target Acquisition and Recognition (MSTAR) data set. When the number of training samples is 14.5% of the original training set, the recognition rate of the ATR model still reaches 91.27% with the help of the proposed approach.

Mainstream CNN-based SAR ship detectors are prone to produce false targets in the land area, which has a great relationship with the anchor generation mechanism in the network structure. The anchor generator will indiscriminately generate a set of anchors for each point on the predicted feature map and map it back to the original image. These anchors will become the initial candidate boxes after screening. Considering that ship targets in SAR images are usually small and sparsely distributed, this dense anchor generation method can easily generate candidate boxes in land areas. To solve this problem, this paper proposes a method to guide the generation of anchors through image area information, called selective anchor generator (SAG). Unlike previous anchor generator, SAG will do sea-land segmentation for image first and then establish a mapping relationship with the feature map. Points on the feature map that only contain land feature information are marked as negative points with no anchors being generated. In this way, the number of candidate prediction boxes in the land area is greatly reduced, thereby achieving the effect of false alarm suppression. Experiments on test data show that this method can effectively reduce the number of false targets generated in land areas.

Deep convolutional neural networks (DCNNs) have been widely used for SAR image target recognition. However, the huge demands of DCNNs for computing, storage, and energy resources limit their use on edge computing devices. In this article, we propose a method based on image quality evaluation (IQE) criterion to prune deep neural networks. We use IQE criterion to identify unimportant filters, and then remove them, to obtain a lightweight network while maintaining the performance of the neural network as much as possible. Besides, we verified the effectiveness of our method on the MSTAR dataset with cheap edge computing devices.

Manually labeling SAR images is an extremely important but expensive task. Clustering SAR images without labels can effectively reduce the cost of labeling. The existing methods often fail to obtain the optimal solution because the combination of feature representation and clustering assignment, as well as the mining of various correlations behind the input image are ignored. To tackle these problems, an end-to-end deep clustering framework based on a dynamic threshold for SAR images is proposed. In this method, clustering is transformed into a binary pairwise classification problem, and the positive and negative image pairs with high confidence are selected gradually through a dynamic threshold to guide the network representation learning. Various correlations of images are mined and used to learn the most comprehensive representation by reconstructing and maximizing the mutual information between the deep and shallow layers of the network. Extensive experiments on MSTAR datasets show that this method achieves better clustering performance than some state-of-the-art methods.

In SAR images, there are objects with large scale difference, which is called cross-scale objects. For example, there are both large-scale airport objects and small-scale ship objects in SAR images. However, the current multi-scale object detection methods are difficult to detect objects with large scale difference. To address this issue, we propose a cross-scale object detection method for SAR images based on Scale Expansion Pyramid Network(SEPN) in this paper. The proposed SEPN can extract the salient features of the objects with a large scale difference, and by closely connecting the scale expansion layer with the convolutional layer during the downsampling process of the Feature Pyramid Network (FPN), the receptive field of the feature extracted by the convolutional layer can be adaptively extended, and finally it achieves the effect of cross-scale object detection in SAR image. Experiments on SSDD dataset and Gaofen-3 dataset show the effectiveness of our proposed methods in detecting objects of different scales in different scenes of SAR images.

The constant false-alarm-rate (CFAR) target detection is an important research direction for synthetic aperture radar (SAR) image application. The traditional pixel-level CFAR method has great shortcomings in eliminating the false-alarm targets and keeping the complete information of the targets. Thus, the superpixel-level CFAR has become an important topic in research in recent years. However, the current superpixel-level CFAR methods have not considered or built a superpixel clutter-distribution model. Therefore, an improved CFAR based on superpixel modeling is proposed in this letter. A superpixel-level compound Gamma distribution was built to describe the clutter statistical model, which can obtain a more accurate fitting than the pixel-level Gamma distribution. The experiments on the SAR images verified that the proposed method can effectively suppress the influence of the background clutter to reduce the false alarms and can keep the complete shape information of the targets. As a result, the proposed method outperforms the traditional pixel-level CFAR and the current superpixel-level CFAR methods.

Ship target detection using large-scale synthetic aperture radar (SAR) images has important application in military and civilian fields. However, ship targets are difficult to distinguish from the surrounding background and many false alarms can occur due to the influence of land area. False alarms always occur with ship target detection because most of the area in large-scale SAR images are treated as background and clutter, and the ship targets are considered unevenly distributing small targets. To address these issues, a ship detection method in large-scale SAR images via CenterNet is proposed in this article. As an anchor-free method, CenterNet defines the target as a point, and the center point of the target is located through key point estimation, which can effectively avoid the missing detection of small targets. At the same time, the spatial shuffle-group enhance (SSE) attention module is introduced into CenterNet. Through SSE, the stronger semantic features are extracted while suppressing some noise to reduce false positives caused by inshore and inland interferences. The experiments on the public SAR-ship-data set show that the proposed method can detect all targets without missed detection with dense-docking targets. For the ship targets in large-scale SAR images from Sentinel 1, the proposed method can also detect targets near the shore and in the sea with high accuracy, which outperforms the methods like faster R-convolutional neural network (CNN), single-shot multibox detector (SSD), you only look once (YOLO), feature pyramid network (FPN), and their variations.

Ship target interpretation in SAR images has become an important topic in research in recent years. With the improvement of SAR image resolution, the performance of traditional automatic target recognition (ATR) method decreases gradually. The emergence of deep network provides a new solution for SAR image ship interpretation. Ship target interpretation in SAR images based on deep learning is divided into detection and classification, but they haven't been integrated yet. Based on the process flow of traditional ATR system, an integration method of ship target detection and recognition based on deep network is proposed in this paper. And at the end of the network, the squeeze-and-excitation (SE) module is added to the classification subnetwork. The effectiveness of the proposed integration method is verified by experiments, and the classification accuracy of the ship increased by 3.7% after adding SE module.

Target detection for dense docked SAR ships has always been a challenge. First of all, dense docked ships are generally in the port area, and the interference in the land area is large. Secondly, the adjacent ships are easily detected as a ship in the detection, or will be suppressed during the Non Maximal Suppression (NMS) process, causing the targets to be lost. This paper proposes a target detection method for dense docked ships based on CenterNet. At the same time, Spatial Group-wise Enhance (SGE) attention module is added to CenterNet in this paper. SGE reduces the amount of calculation by grouping channels, and at the same time strengthens the spatial features of each group to extract more semantic features. The enhanced feature map is sent to the detection network for dense docked SAR ship target detection. The proposed method is verified on the dataset SAR-ship-Dataset, and the experimental results show that the method in this paper has better detection performance for dense docked ships.

With the development of imaging technology, ship target detection in large scenes has become a research hotspot. The patches without sea area sent to detector greatly increase the computational cost and there are many false alarms in land area. Based on above, this paper proposes a ship detection method based on bias classification. Patches of large scale SAR images without sea area will no longer be sent to the detector, which greatly reduces false alarms in land area. The proposed method is implemented in the anchor-free network framework named CenterNet. The experimental results show that the bias classification method proposed in this paper can effectively reduce false alarms in land area.

Gestures can be gradually used to achieve natural and direct communication between people and machines, not limited to people. However, micro gesture motion sensing using traditional sensors and techniques is challenging because of the difference and diversity of finger motion. This paper proposes a novel method for micro gesture recognition based on the high resolution of terahertz radar, which can capture fine changes during gesture movement. From these radar echoes collected in gesture motion, the low dimensional projection features of diagonal profile based on range-Doppler map (RDM) are extracted to characterize the motion difference between gestures. Then, a convolutional neural network (CNN) is employed to realize the gesture recognition system. The experimental results show that the proposed method can achieve a high recognition rate of 98.06% with six dynamic micro gestures across eight participants.

Under the framework of a supervised learning-based automatic target recognition (ATR) approach, recognition performance is primarily dependent on the amount of training samples. However, shortage in training samples is a consistent issue for ATR. In this article, we propose a new image to image generation method, called label-directed generative adversarial networks (LDGANs), which will provide labeled samples to be used for recognition model training. We define an entirely new loss function for the LDGAN, which utilizes the Wasserstein distance to replace the original distance measurement of the conventional generative adversarial networks (GANs), thus efficiently avoiding the collapse mode problem. The label information is also added to the loss function of the LDGAN to avoid generating a large number of unlabeled target images. More importantly, the proposed method also makes corresponding changes to the network architecture regarding the new GANs. At the same time, the detailed algorithm about the LDGAN is also introduced in this article to deal with the issue that characteristically GANs are not easy to train. Based on comparisons with other directed generation methods, the experimental results show comparative results of several types of generated images in statistical features, gradient features, classic features of synthetic aperture radar (SAR) targets and the independence from the real image. While demonstrating that the images generated by the LDGAN produced better results using the assumptions of independent and identical distribution, the experiment also explores the performance of the generated image in the ATR. A comparison of these experimental results demonstrates a better way to use the generated image for ATR. The experimental results also prove that the proposed method does have the ability to supplement information for ATR when the training sample information is insufficient.

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.

Noncontact gesture recognition is gradually being applied to emerging applications, such as smart cars and smart phones. Negative latency gesture recognition (recognition before a gesture is finished) is desirable due to the instantaneous feedback. However, it is difficult for existing methods to achieve a high precision and negative latency gesture recognition. A fragment can provide too few features to directly identify all gestures well. By observing a large number of existing gesture sets and people's daily operating habits, we found that some high frequency used gestures are similar. To the best of our knowledge, it is the first time to redivide the gestures into two subsets according to their movement physical states. We divided the gestures with different shapes or motion states into a parent-class subset, and further divided each pair of parent-class gestures to obtain a child-class subset. In order to achieve a better tradeoff between the high-precision and negative latency, an approach of motion pattern and behavior intention (MPBI) is proposed. Taking full advantage of the characteristics of each subset, MPBI includes two models. First, pattern model coarsely classify the parent-class gestures by a convolutional network, and then intention model further classifies child-class gestures according to their opposite motion direction. MPBI is evaluated on a 340-GHz terahertz radar. With the advantage of its accurate ranging, intention model can recognize child-class gestures directly without training. MPBI is evaluated on 12 gestures and achieves a recognition accuracy of 94.13%, which only needs a 0.033-s gesture fragment as an input sample.

Previous convolutional neural networks (CNNs) used for synthetic aperture radar (SAR) target recognition are over-parameterized which limits their application in real-time radar recognition systems. To solve this problem, a micro convolution neural network (MCNN) for SAR target recognition is proposed in this paper. Our MCNN is compressed from a deep convolutional neural network (DCNN) with 18 layers by a novel knowledge distillation algorithm. The experiments on MSTAR dataset show that the proposed MCNN can obtain the recognition rate of 98.2%. This recognition rate is almost the same as the DCNN. However, compared with the DCNN, the memory footprint of the proposed MCNN is compressed by nearly 177 times, and the calculated amount is nearly 12.8 times less, which means that the proposed MCNN can obtain a better performance with the smaller network.

The scales of different ships vary in synthetic aperture radar (SAR) images, especially for small scale ships, which only occupy few pixels. So ship detection methods currently face difficulties in detecting multiscale ships. A novel method for multiscale ship detection in SAR images based on Dense Attention Pyramid Network (DAPN) is proposed in this paper. It can extract multiscale and salient features by DAPN, which densely connects Convolutional Block Attention Module (CBAM) to each feature map from top to down of the pyramid network. Then the fused feature maps are fed to the detection network for multiscale ship detection. Experiments on SSDD dataset show a better performance of this method to detect multiscale ships in different scenes of SAR images.

The scales of different ships vary in synthetic aperture radar (SAR) images, especially for small scale ships, which only occupy few pixels. So ship detection methods currently face difficulties in detecting multiscale ships. A novel method for multiscale ship detection in SAR images based on Dense Attention Pyramid Network (DAPN) is proposed in this paper. It can extract multiscale and salient features by DAPN, which densely connects Convolutional Block Attention Module (CBAM) to each feature map from top to down of the pyramid network. Then the fused feature maps are fed to the detection network for multiscale ship detection. Experiments on SSDD dataset show a better performance of this method to detect multiscale ships in different scenes of SAR images.

Previous ship detection methods for synthetic aperture radar (SAR) images suffer from an extreme variance of ship scale. The problem of large scale variation across ships lies in the heart of ship detection. In this paper, scale-transferrable pyramid network for multi-scale ship detection in SAR images is proposed. We construct a feature pyramid network by lateral connection, and densely connect each feature maps from top to down using scale-transfer layer. Lateral connection injects more semantic information into feature maps with high resolution. Dense scale-transfer connection can expand the resolution of feature maps and explicitly explore information contained in channels. Finally, we can detect multi-scale ships by combining these multi-scale feature maps. Experimental results demonstrate that our network outperforms the state-of-the-art methods.

The Synthetic Aperture Radar (SAR) target detection using Adaptive Boosting (AdaBoost) based on Haar-like feature is accelerated via Graphics Processing Unit (GPU) in this paper. As a machine learning algorithm, AdaBoost has achieved great success in the field of target detection. However, due to the time-consuming training process, it is difficult to achieve real time requirements, which limits its further development. In this paper, based on the analysis of the algorithm, the AdaBoost algorithm based on Haar-like feature is parallel decomposed and then implemented by using Moving and Stationary Target Recognition (MSTAR) dataset to improve the detection efficiency. First, the AdaBoost algorithm based on Haar-like feature is investigated. Then, in order to improve the detection speed, the algorithm is parallel decomposed. Finally, the algorithm is implemented by Compute Unified Device Architecture (CUDA) for acceleration to see the acceleration effect. According to the experiments, the time spent on the training process and the testing process has been greatly reduced by using CUDA. Compared to the traditional CPU-based AdaBoost algorithm based on Haar-like feature, the algorithm using CUDA parallel computing can achieve a speedup of 30.

As one of the most commonly used natural languages, gesture has become an important approach for people to interact with machines. However, the flexibility of fingers has posed a challenge for detecting fine-grained gestures. In this paper, a novel method of high precision fine-grained gesture recognition is proposed based on a terahertz radar, which is able to sense any gesture movement when its range of motion is greater than 5mm. First, High Resolution Range Profile (HRRP) sequences are extracted from radar echoes. Then, the HRRP features are fed to Random Forest classifier after dimensionality reduction by PCA. In order to verify the proposed method is effective to detect fine-grained gestures, four kinds of similar gestures with multi-fingers are designed for experiments. The results indicate the recognition rate exceeds 99.7%, which demonstrates a great prospect in fine-grained gesture recognition using a terahertz radar.

Synthetic aperture radar (SAR) is an active microwave imaging sensor with the capability of working in all-weather, all-day to provide high-resolution SAR images. Recently, SAR images have been widely used in civilian and military fields, such as ship detection. The scales of different ships vary in SAR images, especially for small-scale ships, which only occupy few pixels and have lower contrast. Compared with large-scale ships, the current ship detection methods are insensitive to small-scale ships. Therefore, the ship detection methods are facing difficulties with multi-scale ship detection in SAR images. A novel multi-scale ship detection method based on a dense attention pyramid network (DAPN) in SAR images is proposed in this paper. The DAPN adopts a pyramid structure, which densely connects convolutional block attention module (CBAM) to each concatenated feature map from top to bottom of the pyramid network. In this way, abundant features containing resolution and semantic information are extracted for multi-scale ship detection while refining concatenated feature maps to highlight salient features for specific scales by CBAM. Then, the salient features are integrated with global unblurred features to improve accuracy effectively in SAR images. Finally, the fused feature maps are fed to the detection network to obtain the final detection results. Experiments on the data set of SAR ship detection data set (SSDD) including multi-scale ships in various SAR images show that the proposed method can detect multi-scale ships in different scenes of SAR images with extremely high accuracy and outperforms other ship detection methods implemented on SSDD.

Incremental learning methods update the existing model with new knowledge when the target data increase continuously. Open set recognition (OSR) algorithms provide classifiers with a rejection option so that the new untrained target type is identified. In this paper, an open set incremental learning method is introduced for automatic target recognition, which is able to recognize and learn the new unknown classes continually. The proposed method, open set model with incremental learning (OSmIL), is an ensemble classifier so it is able to be updated only by the new data. For saving the computational time and storage source, a new exemplar selection method is introduced for model simplifying. Edge samples are selected to cover training classes; as a result, the model size is deduced and controlled. Moreover, because extreme value theory (EVT) is suitable to fit a classification model that includes open space risk, the decision function based on EVT makes an open set classifier for identifying the new classes. Experimental results demonstrate that the proposed OSmIL outperforms the other state of the arts on the accuracy of multiclass OSR. And OSmIL can maintain good accuracy and efficiency in the incremental learning experiment set.

The detection of airports using synthetic aperture radar (SAR) images has attracted considerable attention. Traditional methods easily result in inaccurate detection due to the complex scenes and multiplicative speckle noise. Therefore, airport detection from SAR images is still a challenging task. In order to limit the influence of unnecessary and attractive details and noise, we propose a multi-layer abstraction saliency model for airport detection in SAR images in this paper. Specifically, we first obtain airport support regions and superpixels in the first layer. According to the dis-similarity between foreground and background superpixels, airport components are explored by iterative refinement for each airport support region in the second layer. In the third layer, airport adobes are produced by clustering. Based on the characteristics of an airport in SAR images, we propose three saliency cues, including local contrast (LC), adobe deformation (AD), and global uniqueness (GU), to obtain adobe-level saliency. Furthermore, we assign saliency to each pixel by Bayesian inference. Finally, we can explore airport location using integrated saliency map. The proposed approach is tested on an airport data set collected from Gaofen-3, TerraSAR, and RadarSat. Our method achieves 88.89% detection rate. The experimental results demonstrate that the proposed algorithm is effective and outperforms the previously airport detection methods. The code will be available at https://github.com/NengyuanLiu/MyAirportSaliency.

A SAR-ATR method with favorable performance on translational invariance is proposed in this paper. Nowadays, supervised learning is the main way to realize image processing and target detection, but performance of most trained models will be greatly influenced by distribution of targets in training samples. In some recent researches of SAR-ATR, models trained on MSTAR dataset apply methods of data augmentation to increase randomness of target position in training samples, so that the models could have a certain capability of translational invariance. The method that we proposed based on convolutional neural network (CNN) doesn't need operation of data augmentation at all, and the comparison experiments based on manually shifted target slices proved that our model performed better in detecting and recognizing shifted targets. Experiments show that the proposed method can accurately locate position of different kinds of shifted targets and realize detection and recognition correctly.

Deep Convolutional Neural Networks (DCNNs) have been widely used in target recognition due to the availability of large dataset. The DCNNs have the ability of learning highly hierarchical image feature, which provides great opportunity for synthetic aperture radar automatic target recognition (SAR-ATR). However, when the DCNNs were directly applied to the SAR target recognition, it will result in severe overfitting due to limited SAR image training data. To overcome this problem, we present a Gabor-Deep Convolutional Neural Networks (G-DCNNs). Instead of training a deep network with limited dataset of raw SAR images, Gabor features for multi -scale and multi -direction were used for data augmentation as training dataset at first. Then based on this data augmentation method, we designed a DCNNs for SAR image target recognition. Experimental results on the Moving and Stationary Target Acquisition and Recognition (MSTAR) dataset prove the effectiveness of our method.

Directly automatic target recognition (D-ATR) for large scene SAR images based on a deep neural network is proposed in this paper. To recognize the targets in SAR images, the traditional methods contains four steps: detection, discrimination, feature extraction and classification. These processes are independent but the processing result from the previous step will affect the next step, and there is still room for processing speed improvement of the latest integrated system of SAR target detection. The proposed method can integrate these steps as a whole system and directly recognize targets in large scene SAR images by encapsulating all computation in a single deep convolutional neural network (DCNN). Experiments on MSTAR dataset and the 1478×1784 simulated large scene SAR image show high accuracy with a fast processing speed.

Large-scale high-quality, standardized, measurable and accurate data is the key to promote the progress of the algorithm in the radar remote sensing. Data scaling is a widespread technology that increases the size of a labeled training set dataset through specific data transformations. Synthetic Aperture Radar (SAR) image simulators based on computer-aided mapping models play an important role in SAR applications such as automatic target recognition and image interpretation, but the accuracy of this simulator is due to geometric errors and simplification of electromagnetic calculations. In order to achieve a SAR image datasets with the known target and azimuth angles, we can generate the desired image directly from a known image database. We can realize the augmentation of SAR image data set through linear synthesis and Generative Adversarial Networks, which can generate SAR images for the specified azimuth.

Compared with optical images, the number of synthetic aperture radar (SAR) images is limited and growing slowly over time, and most of time the additional SAR images are unlabeled, which restricts the development of target recognition on SAR images. In this letter, the update learning method for SAR unlabeled target recognition based on the convolutional neural network (CNN) with assistant decision is proposed. The seed image set, which contains a small amount of existing labeled SAR images, is used for pretraining a CNN model. The unlabeled samples are fed to both CNN and the assistant classifier, and then, the matrices of the unlabeled samples based on recognition probability will be obtained, respectively. The unlabeled samples with high confidence by decision will be fed to CNN again to fine-tune the CNN model. The proposed method can make full use of the unlabeled samples to update the recognition model and to improve the recognition performance. Experiments on MSTAR data set show that, with the increase of unlabeled samples, the proposed method can improve the recognition performance and outperform the methods of CNN self-update learning and support vector machine update learning with the principal component analysis and the nonnegative matrix factorization. The experiments’ results demonstrate that the proposed method can effectively solve the problems caused by the lack of training samples.

The detection of airports using synthetic aperture radar (SAR) images has attracted considerable attention. Traditional methods locate airports by connecting pairs of line segments or directly applying saliency analysis to an entire SAR image. These methods are either time-consuming or can easily result in false detection. Considering these issues, a method using line segment grouping and saliency analysis is proposed in this letter. First, line segments are obtained via an improved line segment detector (LSD). After line segment grouping, airport support regions are extracted. Then, selective nonmaximum suppression is proposed to obtain potential airport regions. Finally, airport regions are located by false alarm control and saliency analysis. Experiments on large-scale SAR images prove that our proposed algorithm has a better performance and higher efficiency in airport detection compared with traditional methods.

The target detection and recognition integration in large scene SAR images based on Region Proposal Regression (RPR) is proposed in this paper. In traditional three-stage process of SAR target recognition: detection-discrimination-classification/recognition, many factors between detection and recognition will greatly affect the result of recognition, such as the difference of target region size and target location between detection results and training samples. The proposed method which uses the structure of Deep Convolutional Neural Network (DCNN), can integrate the traditional three-stage process as a whole system at the base of detecting and recognizing targets at the same time. The experiments based on SAR simulation image data show that, the proposed method can accurately and directly recognize multi-class SAR targets in large scene images.