Submit Manuscript
Peer Review
Editor Work2018 Vol. 7, No. 5
Previous Issue
2018, 7(5): 531-547.
Abstract
3583KB
The technique of radar feature extraction, imaging, and recognition of target with micro-motions has become one of the most potential research directions in the field of radar target accurate recognition. In this paper, the concept of micro-motion is first introduced briefly. Subsequently, the achievements of echo modeling, feature extraction, imaging, and identification of micro-motion targets are summarized. Several typical frontier applications are then introduced. Finally, the future development trends of the research are discussed.
The technique of radar feature extraction, imaging, and recognition of target with micro-motions has become one of the most potential research directions in the field of radar target accurate recognition. In this paper, the concept of micro-motion is first introduced briefly. Subsequently, the achievements of echo modeling, feature extraction, imaging, and identification of micro-motion targets are summarized. Several typical frontier applications are then introduced. Finally, the future development trends of the research are discussed.
2018, 7(5): 548-556.
Micromotion refers to the small and non-uniform motion of the target or several target components along the radar line of sight. Using the high-resolution three-Dimensional (3D) Inverse Synthetic Aperture Radar (ISAR) imaging, the structural information and motion status of micromotion targets can be obtained, providing essential features for the detection, tracking, identification, and classification, which play important roles in the space situation awareness and ballistic missile defense. Given the complex micromotion forms and the non-stationary radar echoes, the available parametric ISAR imaging methods are no longer applicable. To overcome this limitation, this study aims to propose a high-resolution 3D imaging method for micromotion targets based on the scattering center trajectory matrix decomposition. First, the Range Instantaneous Doppler (RID) image series is generated to extract the support region of scattering centers by the watershed method. Then, the scattering center association is achieved based on the minimum Euclidean distance criterion. Considering the insufficient accuracy in the instantaneous slant range estimation with limited range resolution, a method for refined estimation of the trajectory matrix based on the modern spectrum analysis is proposed. Finally, the high-resolution 3D imaging of the micromotion targets is obtained by the trajectory matrix decomposition with constraints. The simulation results have demonstrated that the proposed method could effectively obtain high-resolution 3D imaging of the targets in complex micromotions such as nutation.
Micromotion refers to the small and non-uniform motion of the target or several target components along the radar line of sight. Using the high-resolution three-Dimensional (3D) Inverse Synthetic Aperture Radar (ISAR) imaging, the structural information and motion status of micromotion targets can be obtained, providing essential features for the detection, tracking, identification, and classification, which play important roles in the space situation awareness and ballistic missile defense. Given the complex micromotion forms and the non-stationary radar echoes, the available parametric ISAR imaging methods are no longer applicable. To overcome this limitation, this study aims to propose a high-resolution 3D imaging method for micromotion targets based on the scattering center trajectory matrix decomposition. First, the Range Instantaneous Doppler (RID) image series is generated to extract the support region of scattering centers by the watershed method. Then, the scattering center association is achieved based on the minimum Euclidean distance criterion. Considering the insufficient accuracy in the instantaneous slant range estimation with limited range resolution, a method for refined estimation of the trajectory matrix based on the modern spectrum analysis is proposed. Finally, the high-resolution 3D imaging of the micromotion targets is obtained by the trajectory matrix decomposition with constraints. The simulation results have demonstrated that the proposed method could effectively obtain high-resolution 3D imaging of the targets in complex micromotions such as nutation.
2018, 7(5): 557-564.
Classification of drones is important due to their increasing popularity and potential threats. The micro-Doppler signatures that depend on the rotation of rotor blades facilitate the classification of drones. To enhance the robustness of micro-Doppler based classification of drones, dual radar sensing classification scheme is proposed in this paper. First, time-frequency spectrograms are obtained by performing a short-time Fourier transform on the radar data collected by two radar sensors that have similar angular diversity. Then, principal components analysis is utilized to extract the features from the time-frequency spectrograms and the features obtained by the two radar sensors are fused together. Finally, the classification results are obtained by using the support vector machine. The experimental results show that the classification accuracy obtained by the fusion of dual radar sensors is 5% higher than that obtained by only using a single radar sensor.
Classification of drones is important due to their increasing popularity and potential threats. The micro-Doppler signatures that depend on the rotation of rotor blades facilitate the classification of drones. To enhance the robustness of micro-Doppler based classification of drones, dual radar sensing classification scheme is proposed in this paper. First, time-frequency spectrograms are obtained by performing a short-time Fourier transform on the radar data collected by two radar sensors that have similar angular diversity. Then, principal components analysis is utilized to extract the features from the time-frequency spectrograms and the features obtained by the two radar sensors are fused together. Finally, the classification results are obtained by using the support vector machine. The experimental results show that the classification accuracy obtained by the fusion of dual radar sensors is 5% higher than that obtained by only using a single radar sensor.
2018, 7(5): 565-574.
In this paper, Convolutional Neural Networks (CNN) are used to detect and classify micro-Doppler effects of maritime targets by using generalized learning ability for high-dimensional features. Based on the micro-motion model of maritime targets, two-dimensional time-frequency maps of four types of micro-motion signals are constructed in the measured sea clutter background. These maps were used as training and test datasets. Furthermore, three types of CNN models, i.e., LeNet, AlexNet, and GoogleNet, are used in binary detection and multiple micro-motion classifications. The effects of signal-to-noise ratio on detection and classification performance are also studied. Compared with the traditional support vector machine method, the proposed method can learn the micro-motion features intelligently, and has performed better in detection and classification. Thus, this study can provide a new technical approach for radar target detection and recognition under a cluttered background.
In this paper, Convolutional Neural Networks (CNN) are used to detect and classify micro-Doppler effects of maritime targets by using generalized learning ability for high-dimensional features. Based on the micro-motion model of maritime targets, two-dimensional time-frequency maps of four types of micro-motion signals are constructed in the measured sea clutter background. These maps were used as training and test datasets. Furthermore, three types of CNN models, i.e., LeNet, AlexNet, and GoogleNet, are used in binary detection and multiple micro-motion classifications. The effects of signal-to-noise ratio on detection and classification performance are also studied. Compared with the traditional support vector machine method, the proposed method can learn the micro-motion features intelligently, and has performed better in detection and classification. Thus, this study can provide a new technical approach for radar target detection and recognition under a cluttered background.
2018, 7(5): 575-584.
The micro-motion feature is one of the important characteristic information of spatial target recognition. However, the existing multifunctional Multi-Input Multi-Output (MIMO) radar usually has to allocate a large number of continuous time resources for target micro-motion feature extraction after target searching and tracking, which leads to a low real-time performance of target recognition and poor overall performance of radar system. To solve this problem, this paper presents a multi-target micro-motion feature extraction method for MIMO radar based on tracking pulses. First, according to the azimuth information of each target, the MIMO radar transmitting waveform is designed, and the tracking pulses are transmitted simultaneously for targets with different directions. On this basis, by considering the micro-motion feature extraction performance and the target tracking performance synthetically, the transmission time series of the tracking pulses are optimized. Finally, the narrowband tracking pulses are directly used to simultaneously extract the micro-motion features of the targets in different directions, which makes it no longer necessary to allocate additional radar resources for target feature extraction. Consequently, the real-time recognition performance and the working efficiency of radar are improved significantly. Simulations demonstrate that when the signal-to-noise ratio is larger than –10 dB, the micro-motion features of multi-targets can be extracted accurately, which verifies the effectiveness and robustness of the proposed method.
The micro-motion feature is one of the important characteristic information of spatial target recognition. However, the existing multifunctional Multi-Input Multi-Output (MIMO) radar usually has to allocate a large number of continuous time resources for target micro-motion feature extraction after target searching and tracking, which leads to a low real-time performance of target recognition and poor overall performance of radar system. To solve this problem, this paper presents a multi-target micro-motion feature extraction method for MIMO radar based on tracking pulses. First, according to the azimuth information of each target, the MIMO radar transmitting waveform is designed, and the tracking pulses are transmitted simultaneously for targets with different directions. On this basis, by considering the micro-motion feature extraction performance and the target tracking performance synthetically, the transmission time series of the tracking pulses are optimized. Finally, the narrowband tracking pulses are directly used to simultaneously extract the micro-motion features of the targets in different directions, which makes it no longer necessary to allocate additional radar resources for target feature extraction. Consequently, the real-time recognition performance and the working efficiency of radar are improved significantly. Simulations demonstrate that when the signal-to-noise ratio is larger than –10 dB, the micro-motion features of multi-targets can be extracted accurately, which verifies the effectiveness and robustness of the proposed method.
2018, 7(5): 585-592.
In this paper, experimental results of micro-Doppler effect on a multi-rotor drone with digital television based passive radar are discussed. First, the bistatic passive radar micro-motion model of the drone is established. Second, key techniques of micro-motion signal extraction are briefly described. Finally, the experimental process is introduced, including experimental scene configuration and analysis of typical experimental results of micro-Doppler effect. The experimental results agree with the theoretical analysis of the motion parameters of the drone, thereby confirming the technical feasibility of detecting the micro-Doppler effect of the multi-rotor drone using the digital television based passive radar.
In this paper, experimental results of micro-Doppler effect on a multi-rotor drone with digital television based passive radar are discussed. First, the bistatic passive radar micro-motion model of the drone is established. Second, key techniques of micro-motion signal extraction are briefly described. Finally, the experimental process is introduced, including experimental scene configuration and analysis of typical experimental results of micro-Doppler effect. The experimental results agree with the theoretical analysis of the motion parameters of the drone, thereby confirming the technical feasibility of detecting the micro-Doppler effect of the multi-rotor drone using the digital television based passive radar.
2018, 7(5): 593-601.
In this paper, a Sinusoidal Frequency Modulation Fourier-Bessel Transform (SFMFBT) is proposed for micro-Doppler (m-D) target feature extraction. Initially, the definition of SFMFBT is given, and then, some of its properties are analyzed. A method to reduce frequency extraction error is then introduced based on error analysis. Finally, some issues in the frequency extraction of discrete signals are discussed. After comparing with the values of Fourier-Bessel Transform (FBT), SFMFBT introduces the parameter k-resolution into the kernel function of FBT so that accuracy of the signal decomposition is distinctly improved. In addition, from the error analysis in SFMFBT, Bessel-based signal decomposition methods can be used for feature extraction, whose application scope is extended. Simulation results verify the effectiveness of the proposed method for a group m-D target and also verifies its robustness when SNR>0 dB.
In this paper, a Sinusoidal Frequency Modulation Fourier-Bessel Transform (SFMFBT) is proposed for micro-Doppler (m-D) target feature extraction. Initially, the definition of SFMFBT is given, and then, some of its properties are analyzed. A method to reduce frequency extraction error is then introduced based on error analysis. Finally, some issues in the frequency extraction of discrete signals are discussed. After comparing with the values of Fourier-Bessel Transform (FBT), SFMFBT introduces the parameter k-resolution into the kernel function of FBT so that accuracy of the signal decomposition is distinctly improved. In addition, from the error analysis in SFMFBT, Bessel-based signal decomposition methods can be used for feature extraction, whose application scope is extended. Simulation results verify the effectiveness of the proposed method for a group m-D target and also verifies its robustness when SNR>0 dB.
2018, 7(5): 602-612.
The integrated Synthetic Aperture Radar (SAR) and Communication system can improve the networking capabilities of SAR, realize the real-time transmission of detection data, and improve the overall performance of the system. In the working process of the integrated platform, Doppler shift and multipath effects will be introduced, which make the orthogonality of the widely studied Orthogonal Frequency Division Multiplexing (OFDM) integrated waveforms unable to be maintained, and the imaging and communication performance is impaired. In this paper, the Filter Bank Multicarrier (FBMC) waveform used for the integrated SAR and communication systems is proposed. On the one hand, FBMC waveforms require low orthogonality between subcarriers, which can counter Doppler and multipath effects. On the other hand, FBMC waveforms do not use Cyclic Prefix (CP), so the false targets can be avoided and spectral efficiency can be improved. This paper analyzes the integrated performance of FBMC waveforms, studies the effects of multipath and Doppler shift on FBMC waveforms in the integrated system, and proposes a Doppler compensation algorithm for integrated FBMC waveform with large frequency shift. According to the above analysis, the FBMC waveform has better performance in the wide swath SAR and communication integration system. Simulation results verify the conclusion.
The integrated Synthetic Aperture Radar (SAR) and Communication system can improve the networking capabilities of SAR, realize the real-time transmission of detection data, and improve the overall performance of the system. In the working process of the integrated platform, Doppler shift and multipath effects will be introduced, which make the orthogonality of the widely studied Orthogonal Frequency Division Multiplexing (OFDM) integrated waveforms unable to be maintained, and the imaging and communication performance is impaired. In this paper, the Filter Bank Multicarrier (FBMC) waveform used for the integrated SAR and communication systems is proposed. On the one hand, FBMC waveforms require low orthogonality between subcarriers, which can counter Doppler and multipath effects. On the other hand, FBMC waveforms do not use Cyclic Prefix (CP), so the false targets can be avoided and spectral efficiency can be improved. This paper analyzes the integrated performance of FBMC waveforms, studies the effects of multipath and Doppler shift on FBMC waveforms in the integrated system, and proposes a Doppler compensation algorithm for integrated FBMC waveform with large frequency shift. According to the above analysis, the FBMC waveform has better performance in the wide swath SAR and communication integration system. Simulation results verify the conclusion.
2018, 7(5): 613-621.
Radar target classification is very important in military and civilian fields. Extreme Learning Machines (ELMs) are widely used in classification because of their fast learning speed and good generalization performance. However, because of their shallow architecture, ELMs may not effectively capture the data high level abstractions. Although many researchers have proposed the Deep Extreme Learning Machine (DELM), which can be used to automatically learn high level feature representations, the model easily falls into overfitting when the training sample is limited. To address this issue, Dropout Constrained Deep Extreme Learning Machine (DCDELM) is proposed in this paper. The experimental results on the measured radar data show that the accuracy of the proposed algorithm can reach 93.37%, which is 5.25% higher than that of the stacked autoencoder algorithm, and 8.16% higher than that of the traditional DELM algorithm.
Radar target classification is very important in military and civilian fields. Extreme Learning Machines (ELMs) are widely used in classification because of their fast learning speed and good generalization performance. However, because of their shallow architecture, ELMs may not effectively capture the data high level abstractions. Although many researchers have proposed the Deep Extreme Learning Machine (DELM), which can be used to automatically learn high level feature representations, the model easily falls into overfitting when the training sample is limited. To address this issue, Dropout Constrained Deep Extreme Learning Machine (DCDELM) is proposed in this paper. The experimental results on the measured radar data show that the accuracy of the proposed algorithm can reach 93.37%, which is 5.25% higher than that of the stacked autoencoder algorithm, and 8.16% higher than that of the traditional DELM algorithm.
2018, 7(5): 622-631.
In recent years, Synthetic Aperture Radar (SAR) image target recognition based on the Convolutional Neural Network (CNN) has attracted a significant amount of attention. Fully CNN (FCNN) is a structural improvement of the CNN, which features a higher recognition rate than CNN, but it still requires a large number of labeled data in the training process. This study aims to propose a method of SAR image target recognition based on FCNN and Improved Convolutional Auto-Encoder (ICAE), where several parameters of FCNN are initialized by the parameters of the ICAE encoder. These parameters are obtained in the unsupervised training mode. Then, the FCNN is trained by the labeled training samples. The experimental results on 10 kinds of target classification based on the MSTAR datasets show that this method cannot only achieve an average of 98.14% correct recognition rate but also feature a strong anti-noise capability when the labeled training samples are unexpanded.
In recent years, Synthetic Aperture Radar (SAR) image target recognition based on the Convolutional Neural Network (CNN) has attracted a significant amount of attention. Fully CNN (FCNN) is a structural improvement of the CNN, which features a higher recognition rate than CNN, but it still requires a large number of labeled data in the training process. This study aims to propose a method of SAR image target recognition based on FCNN and Improved Convolutional Auto-Encoder (ICAE), where several parameters of FCNN are initialized by the parameters of the ICAE encoder. These parameters are obtained in the unsupervised training mode. Then, the FCNN is trained by the labeled training samples. The experimental results on 10 kinds of target classification based on the MSTAR datasets show that this method cannot only achieve an average of 98.14% correct recognition rate but also feature a strong anti-noise capability when the labeled training samples are unexpanded.
微信 | 公众平台 
