Junkui Tang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Junkui Tang has a total of 7 co-authors. Here are the top co-authors by number of shared publications:
- Lei Ran
National Key Laboratory of Radar Signal Processing, Xidian University, Xian, China
4 shared publications Rong Xie
National Key Laboratory of Radar Signal Processing, Xidian University, Xian, China
4 shared publications- Zheng Liu
National Key Laboratory of Radar Signal Processing, Xidian University, Xian, China
4 shared publications - Jikai Qin
Air Force Engineering University, Xi'an, Shaanxi, CHINA
3 shared publications - Genquan Han
Xidian University
1 shared publication - Yan Liu
Hangzhou Institute of Technology, Xidian University, Hangzhou 311200, China
1 shared publication
Latest Publications
- 2 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 63 (01962892) , 1-14
Multichannel Radar Forward-Looking Super-Resolution Imaging Method Based on Structured Sparsity
The compressed sensing (CS) imaging method, based on the sparsity of the target in the imaging scene, provides feasibility for achieving radar forward-looking super-resolution imaging. However, traditional CS imaging methods only consider isolated, strong scattering points in the generated image, limiting the quality of forward-looking imaging. To address this problem, this paper proposes a multi-channel radar structured sparse forward-looking super-resolution imaging method based on convolutional weighting. Inspiration for this method stems from the distribution characteristics of scatterers in radar images, where their correlation with neighboring points reflects the intrinsic structural information of the target. By incorporating a convolutional weighting mechanism, the imaging model is able to more effectively extract such structured sparsity key information within the scene. Specifically, when applying the Alternating Direction Multiplier Method (ADMM) to imaging problems, this method dynamically adjusts the weight of the next iteration by analyzing the convolutional relationship between each pixel in the current solution and its neighborhood values, thereby effectively capturing and utilizing the structured sparsity characteristics in the image. Finally, simulation and measured results show that the proposed method not only improves radar forward-looking super-resolution imaging performance but also preserves the structural information of the reconstructed target, proving its effectiveness and superiority.
- 2 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 63 (01962892) , 1-14
Multichannel Radar Forward-Looking Super-Resolution Imaging Method Based on Structured Sparsity
The compressed sensing (CS) imaging method, based on the sparsity of the target in the imaging scene, provides feasibility for achieving radar forward-looking super-resolution imaging. However, traditional CS imaging methods only consider isolated, strong scattering points in the generated image, limiting the quality of forward-looking imaging. To address this problem, this paper proposes a multi-channel radar structured sparse forward-looking super-resolution imaging method based on convolutional weighting. Inspiration for this method stems from the distribution characteristics of scatterers in radar images, where their correlation with neighboring points reflects the intrinsic structural information of the target. By incorporating a convolutional weighting mechanism, the imaging model is able to more effectively extract such structured sparsity key information within the scene. Specifically, when applying the Alternating Direction Multiplier Method (ADMM) to imaging problems, this method dynamically adjusts the weight of the next iteration by analyzing the convolutional relationship between each pixel in the current solution and its neighborhood values, thereby effectively capturing and utilizing the structured sparsity characteristics in the image. Finally, simulation and measured results show that the proposed method not only improves radar forward-looking super-resolution imaging performance but also preserves the structural information of the reconstructed target, proving its effectiveness and superiority.
- 16 January 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 61 (01962892) , 1-12
Enhancing Forward-Looking Image Resolution: Combining Low-Rank and Sparsity Priors
Compressed sensing (CS) based imaging technology has attracted a lot of interest because it can enhance imaging resolution. Targets of interest for forward-looking imaging radar are typically few in comparison to the entire imaging region. This sparsity allows for the natural application of CS to the reconstruction of high-resolution forward-looking images. However, conventional CS-based imaging methods can only perform well when the signal-to-noise ratio (SNR) is high. Strong noise in radar imaging prevents CS-based methods from producing excellent imaging results. Inspired by the low-rank propertity of the received radar target echo and the sparsity of the forward-looking image targets, we present a combined low-rank and sparse prior restricted model for forward-looking imaging with a multi-channel array radar to overcome strong noise. In order to solve the low-rank joint sparse double prior constraint optimization problem, an augmented Lagrange multiplier (ALM) reconstruction method under the framework of the alternating direction multiplier method (ADMM) is proposed. Finally, the results of simulation and real measurement data indicate that our presented method is fairly effective at enhancing the azimuth resolution and robustness of forward-looking radar imaging in comparison to other current methods.
- 16 January 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 61 (01962892) , 1-12
Enhancing Forward-Looking Image Resolution: Combining Low-Rank and Sparsity Priors
Compressed sensing (CS) based imaging technology has attracted a lot of interest because it can enhance imaging resolution. Targets of interest for forward-looking imaging radar are typically few in comparison to the entire imaging region. This sparsity allows for the natural application of CS to the reconstruction of high-resolution forward-looking images. However, conventional CS-based imaging methods can only perform well when the signal-to-noise ratio (SNR) is high. Strong noise in radar imaging prevents CS-based methods from producing excellent imaging results. Inspired by the low-rank propertity of the received radar target echo and the sparsity of the forward-looking image targets, we present a combined low-rank and sparse prior restricted model for forward-looking imaging with a multi-channel array radar to overcome strong noise. In order to solve the low-rank joint sparse double prior constraint optimization problem, an augmented Lagrange multiplier (ALM) reconstruction method under the framework of the alternating direction multiplier method (ADMM) is proposed. Finally, the results of simulation and real measurement data indicate that our presented method is fairly effective at enhancing the azimuth resolution and robustness of forward-looking radar imaging in comparison to other current methods.
- 19 December 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 23 (3) , 3115-3127
Forward-Looking Super-Resolution Imaging Based on Echo Denoising and Noise Weighting at Low SNR
In forward-looking imaging radar sensors, the targets of our focus are typically sparse in comparison to the whole imaging scene, which provides the feasibility of applying compressed sensing (CS) to forward-looking super-resolution imaging. However, sparsity is not satisfied in radar imaging when there is strong noise. This will result in the emergence of false targets and affect imaging results when traditional CS-based approaches are employed directly. In this paper, we propose a noise robust imaging method that combines echo denoising and noise weighting to solve this problem. Firstly, the denoised radar echo is generated using a complex variational mode decomposition (CVMD) method. The CVMD method is an extension of variational mode decomposition (VMD), which has been used to the traditional real signal denoising. According to the signal spectrum characteristics, we successfully applied the proposed CVMD method to radar echo denoising. Secondly, a noise weighting method based on l 1 -norm is performed to further overcome strong noise. The weight comes from the CVMD denoising process, and it can better distinguish targets from the support area of noise. Lastly, simulated and measured data are utilized to verify the proposed method’s performance.
- 19 December 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 23 (3) , 3115-3127
Forward-Looking Super-Resolution Imaging Based on Echo Denoising and Noise Weighting at Low SNR
In forward-looking imaging radar sensors, the targets of our focus are typically sparse in comparison to the whole imaging scene, which provides the feasibility of applying compressed sensing (CS) to forward-looking super-resolution imaging. However, sparsity is not satisfied in radar imaging when there is strong noise. This will result in the emergence of false targets and affect imaging results when traditional CS-based approaches are employed directly. In this paper, we propose a noise robust imaging method that combines echo denoising and noise weighting to solve this problem. Firstly, the denoised radar echo is generated using a complex variational mode decomposition (CVMD) method. The CVMD method is an extension of variational mode decomposition (VMD), which has been used to the traditional real signal denoising. According to the signal spectrum characteristics, we successfully applied the proposed CVMD method to radar echo denoising. Secondly, a noise weighting method based on l 1 -norm is performed to further overcome strong noise. The weight comes from the CVMD denoising process, and it can better distinguish targets from the support area of noise. Lastly, simulated and measured data are utilized to verify the proposed method’s performance.
- Open Access
- 16 August 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 15 (19391404) , 7153-7170
A Target SAR Image Expansion Method Based on Conditional Wasserstein Deep Convolutional GAN for Automatic Target Recognition
For the automatic target recognition (ATR) based on synthetic aperture radar (SAR) images, enough training data are required to effectively characterize target features and obtain good recognition performance. However, in practical applications, it is difficult to collect sufficient training data. To tackle the limitation, a novel end-to-end expansion method, called conditional Wasserstein deep convolutional generative adversarial network with gradient penalty (CWDCGAN), is proposed to achieve SAR image expansion with specified category. To be specific, the CWDCGAN innovatively designed a generative adversarial network architecture based on convolutional and deconvolution networks to improve the quality of generated images. At the same time, conditional information is introduced to control the categories of generated images, and Wasserstein distance and gradient penalty are used to modify the loss function, which makes the network training more stable. Besides, feature extraction and classifier design in a typical ATR system often rely heavily on subjective expert knowledge, which seriously affects its generalization performance. Therefore, a joint recognition method of Resnet18 and support vector machine (Renset18-SVM) is adopted to improve the generalization capacity and the recognition performance. Experimental results with public measured data show that the CWDCGAN can generate higher quality SAR images, and by feeding expanded data to Renset18-SVM, the recognition accuracy is improved under different proportions of training samples.
- 16 August 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 15 (19391404) , 7153-7170
Open AccessA Target SAR Image Expansion Method Based on Conditional Wasserstein Deep Convolutional GAN for Automatic Target Recognition
For the automatic target recognition (ATR) based on synthetic aperture radar (SAR) images, enough training data are required to effectively characterize target features and obtain good recognition performance. However, in practical applications, it is difficult to collect sufficient training data. To tackle the limitation, a novel end-to-end expansion method, called conditional Wasserstein deep convolutional generative adversarial network with gradient penalty (CWDCGAN), is proposed to achieve SAR image expansion with specified category. To be specific, the CWDCGAN innovatively designed a generative adversarial network architecture based on convolutional and deconvolution networks to improve the quality of generated images. At the same time, conditional information is introduced to control the categories of generated images, and Wasserstein distance and gradient penalty are used to modify the loss function, which makes the network training more stable. Besides, feature extraction and classifier design in a typical ATR system often rely heavily on subjective expert knowledge, which seriously affects its generalization performance. Therefore, a joint recognition method of Resnet18 and support vector machine (Renset18-SVM) is adopted to improve the generalization capacity and the recognition performance. Experimental results with public measured data show that the CWDCGAN can generate higher quality SAR images, and by feeding expanded data to Renset18-SVM, the recognition accuracy is improved under different proportions of training samples.