Changyu Hu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Changyu Hu has a total of 32 co-authors. Here are the top co-authors by number of shared publications:
- Ling Wang
Key Laboratory of Radar Imaging and Microwave Photonics of the Ministry of Education, Nanjing University of Aeronautics and Astronautics, Nanjing, China
5 shared publications - Daiyin Zhu
Key Laboratory of Radar Imaging and Microwave Photonics, Ministry of Education, School of Electronic Information Engineering, Nanjing University of Aeronautics and Astronautics, Nanjing 211106, China
3 shared publications - Otmar Loffeld
Center for Sensor Systems (ZESS), University of Siegen, Siegen, Germany
3 shared publications - Chunfeng Chen
Key Laboratory of Radar Imaging and Microwave Photonics, Ministry of Education, Nanjing University of Aeronautics and Astronautics, Nanjing 210016, China
2 shared publications - Jianjiang Zhou
Nanjing University of Aeronautics and Astronautics
2 shared publications - Liam O' Faolain
Munster Technological University
2 shared publications
Latest Publications
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Aerospace and Electronic Systems
- Vol. 62, 13604-13618
Self-Attention Mechanism-Enhanced Dictionary Learning for Sparse ISAR Imaging
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Aerospace and Electronic Systems
- Vol. 62, 13604-13618
Self-Attention Mechanism-Enhanced Dictionary Learning for Sparse ISAR Imaging
- Open Access
- 2 February 2023
- Published by MDPI AG in Electronics
- Vol. 12 (3) , 757
A Fast Power Lines RCS Calculation Method Combining IEDG with CM-SMWA
The existing methods for calculating electromagnetic scattering can be used to obtain the RCS of power lines. However, these methods do not take advantage of the periodicity of power lines. We propose a fast electromagnetic scattering calculation method combining the integral equation discontinuous Galerkin (IEDG) method and the characteristic modes-Sherman–Morrison–Woodbury algorithm (CM-SMWA) exploiting the power lines with stranded structure. We adopt the IEDG to discretize the electric field integral equation (EFIE) so that the EFIE can deal with non-conformal grids and significantly increase the flexibility of the CM-SMWA. Combing with the periodic property of power lines, the modeling and grid generation shall be carried out within one cycle (stranding) of the power line, and the grids of the rest cycle of the power line can be spliced by translating the grid of the divided sections. The advantage of the proposed method lies in that only the CM of one segment needs to be calculated, and the result can be applied to other segments to avoid repeated calculation of the CMs. The simulation results of the RCS of power lines show that the calculation time of our method is cut down by 50% as compared to the conventional CM-SMWA.
- 2 February 2023
- Published by MDPI AG in Electronics
- Vol. 12 (3) , 757
Open AccessA Fast Power Lines RCS Calculation Method Combining IEDG with CM-SMWA
The existing methods for calculating electromagnetic scattering can be used to obtain the RCS of power lines. However, these methods do not take advantage of the periodicity of power lines. We propose a fast electromagnetic scattering calculation method combining the integral equation discontinuous Galerkin (IEDG) method and the characteristic modes-Sherman–Morrison–Woodbury algorithm (CM-SMWA) exploiting the power lines with stranded structure. We adopt the IEDG to discretize the electric field integral equation (EFIE) so that the EFIE can deal with non-conformal grids and significantly increase the flexibility of the CM-SMWA. Combing with the periodic property of power lines, the modeling and grid generation shall be carried out within one cycle (stranding) of the power line, and the grids of the rest cycle of the power line can be spliced by translating the grid of the divided sections. The advantage of the proposed method lies in that only the CM of one segment needs to be calculated, and the result can be applied to other segments to avoid repeated calculation of the CMs. The simulation results of the RCS of power lines show that the calculation time of our method is cut down by 50% as compared to the conventional CM-SMWA.
- Open Access
- 29 June 2022
- Published by MDPI AG in Electronics
- Vol. 11 (13) , 2051
An Improved RCS Calculation Method for Power Lines Combining Characteristic Mode with SMWA
The radar cross-section (RCS) of power lines has an important significance for detection of the power lines. The method of moments (MoM) can calculate the RCS of power lines. However, the efficiency of the MoM is limited by the time-consuming computing process, as well as the expensive storage overhead. In order to enhance the efficiency and reduce the storage of the RCS calculation of power lines, we propose an RCS calculation method that combines the characteristic mode (CM) with a Sherman–Morrison–Woodbury formula-based algorithm (SMWA), which is referred to as a CM-SMWA. CMs are used as the basis functions for reducing the dimension of the MoM impedance matrix, and the SMWA is applied to directly solve the CMs-reduced matrix equation, which can reduce the computational time and storage. The numerical results demonstrate that the proposed method can obtain the RCS of power lines, with different incident angles and different polarizations, at a higher efficiency. At 35 GHz, compared with the conventional MoM, for a typical LGJ50-8 power line with a length of 0.276 m, the computation time is reduced by 62.4% and the memory occupation is reduced by 96.4%.
- 29 June 2022
- Published by MDPI AG in Electronics
- Vol. 11 (13) , 2051
Open AccessAn Improved RCS Calculation Method for Power Lines Combining Characteristic Mode with SMWA
The radar cross-section (RCS) of power lines has an important significance for detection of the power lines. The method of moments (MoM) can calculate the RCS of power lines. However, the efficiency of the MoM is limited by the time-consuming computing process, as well as the expensive storage overhead. In order to enhance the efficiency and reduce the storage of the RCS calculation of power lines, we propose an RCS calculation method that combines the characteristic mode (CM) with a Sherman–Morrison–Woodbury formula-based algorithm (SMWA), which is referred to as a CM-SMWA. CMs are used as the basis functions for reducing the dimension of the MoM impedance matrix, and the SMWA is applied to directly solve the CMs-reduced matrix equation, which can reduce the computational time and storage. The numerical results demonstrate that the proposed method can obtain the RCS of power lines, with different incident angles and different polarizations, at a higher efficiency. At 35 GHz, compared with the conventional MoM, for a typical LGJ50-8 power line with a length of 0.276 m, the computation time is reduced by 62.4% and the memory occupation is reduced by 96.4%.
- 17 August 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Letters
- Vol. 19 (1545598X) , 1-5
FCNN-Based ISAR Sparse Imaging Exploiting Gate Units and Transfer Learning
In recent years, convolutional neural networks (CNNs) have been successfully applied to inverse synthetic aperture radar (ISAR) sparse imaging because of their powerful ability in feature extraction. However, these CNNs only adopt single path feed-forward architectures and lack paths for directly transmitting original feature representations (OFRs) in shallow layers to reconstruction layers, which limits the complete reconstruction of target shape due to the underutilization of the OFRs that are efficient for recovering target details. Later, fully CNN (FCNN) introduces several skip connections (SKs) to establish the additional ways for directly passing the OFRs to the reconstruction layers. Nevertheless, the transmitted OFRs inevitably include the feature information of artifacts, which usually results the appearance of artifacts in final reconstructed target image. To address this issue, we introduce the gate units to FCNN, and refer to the improved FCNN as G-FCNN. Furthermore, the learnable gate units weight the OFRs transmitted by SKs and autonomously decide how many OFRs are transmitted further. To circumvent the shortage of the real data available for network training, we utilize the transfer learning strategy to guarantee a good performance of the G-FCNN. The imaging results of real data show that the G-FCNN-based imaging method is superior to the existing CNN-based imaging methods.
- 17 August 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Letters
- Vol. 19 (1545598X) , 1-5
FCNN-Based ISAR Sparse Imaging Exploiting Gate Units and Transfer Learning
In recent years, convolutional neural networks (CNNs) have been successfully applied to inverse synthetic aperture radar (ISAR) sparse imaging because of their powerful ability in feature extraction. However, these CNNs only adopt single path feed-forward architectures and lack paths for directly transmitting original feature representations (OFRs) in shallow layers to reconstruction layers, which limits the complete reconstruction of target shape due to the underutilization of the OFRs that are efficient for recovering target details. Later, fully CNN (FCNN) introduces several skip connections (SKs) to establish the additional ways for directly passing the OFRs to the reconstruction layers. Nevertheless, the transmitted OFRs inevitably include the feature information of artifacts, which usually results the appearance of artifacts in final reconstructed target image. To address this issue, we introduce the gate units to FCNN, and refer to the improved FCNN as G-FCNN. Furthermore, the learnable gate units weight the OFRs transmitted by SKs and autonomously decide how many OFRs are transmitted further. To circumvent the shortage of the real data available for network training, we utilize the transfer learning strategy to guarantee a good performance of the G-FCNN. The imaging results of real data show that the G-FCNN-based imaging method is superior to the existing CNN-based imaging methods.
- Open Access
- 17 July 2021
- Published by MDPI AG in Remote Sensing
- Vol. 13 (14) , 2812
Inverse Synthetic Aperture Radar Sparse Imaging Exploiting the Group Dictionary Learning
Sparse imaging relies on sparse representations of the target scenes to be imaged. Predefined dictionaries have long been used to transform radar target scenes into sparse domains, but the performance is limited by the artificially designed or existing transforms, e.g., Fourier transform and wavelet transform, which are not optimal for the target scenes to be sparsified. The dictionary learning (DL) technique has been exploited to obtain sparse transforms optimized jointly with the radar imaging problem. Nevertheless, the DL technique is usually implemented in a manner of patch processing, which ignores the relationship between patches, leading to the omission of some feature information during the learning of the sparse transforms. To capture the feature information of the target scenes more accurately, we adopt image patch group (IPG) instead of patch in DL. The IPG is constructed by the patches with similar structures. DL is performed with respect to each IPG, which is termed as group dictionary learning (GDL). The group oriented sparse representation (GOSR) and target image reconstruction are then jointly optimized by solving a norm minimization problem exploiting GOSR, during which a generalized Gaussian distribution hypothesis of radar image reconstruction error is introduced to make the imaging problem tractable. The imaging results using the real ISAR data show that the GDL-based imaging method outperforms the original DL-based imaging method in both imaging quality and computational speed.- 17 July 2021
- Published by MDPI AG in Remote Sensing
- Vol. 13 (14) , 2812
Open AccessInverse Synthetic Aperture Radar Sparse Imaging Exploiting the Group Dictionary Learning
Sparse imaging relies on sparse representations of the target scenes to be imaged. Predefined dictionaries have long been used to transform radar target scenes into sparse domains, but the performance is limited by the artificially designed or existing transforms, e.g., Fourier transform and wavelet transform, which are not optimal for the target scenes to be sparsified. The dictionary learning (DL) technique has been exploited to obtain sparse transforms optimized jointly with the radar imaging problem. Nevertheless, the DL technique is usually implemented in a manner of patch processing, which ignores the relationship between patches, leading to the omission of some feature information during the learning of the sparse transforms. To capture the feature information of the target scenes more accurately, we adopt image patch group (IPG) instead of patch in DL. The IPG is constructed by the patches with similar structures. DL is performed with respect to each IPG, which is termed as group dictionary learning (GDL). The group oriented sparse representation (GOSR) and target image reconstruction are then jointly optimized by solving a norm minimization problem exploiting GOSR, during which a generalized Gaussian distribution hypothesis of radar image reconstruction error is introduced to make the imaging problem tractable. The imaging results using the real ISAR data show that the GDL-based imaging method outperforms the original DL-based imaging method in both imaging quality and computational speed. - 1 December 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-2
Coupled Photonic Crystal Cavity Architecture for Intensity Modulation
Here, we report a novel approach for the intensity modulation of an external cavity laser based on coupled silicon photonic crystal cavities. The cavity resonance was detuned via integrated p-n junction and > 5 dB extinction ratio in transmission was obtained.
- 1 December 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-2
Coupled Photonic Crystal Cavity Architecture for Intensity Modulation
Here, we report a novel approach for the intensity modulation of an external cavity laser based on coupled silicon photonic crystal cavities. The cavity resonance was detuned via integrated p-n junction and > 5 dB extinction ratio in transmission was obtained.
- 9 October 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Letters
- Vol. 17 (7) , 1203-1207
Inverse Synthetic Aperture Radar Imaging Using a Fully Convolutional Neural Network
The traditional inverse synthetic aperture radar (ISAR) imaging uses the range-Doppler (RD) type of methods. The compressive sensing (CS)-based ISAR imaging is capable of obtaining good target images of high contrast and less sidelobe with much less downsampling data. However, the real application of CS ISAR imaging is limited by the time-consuming iteration-based image reconstruction. The image quality is also limited by the performance of sparse representation of the target scene. In recent years, deep learning methods, more specifically the convolutional neural network (CNN), has shown its capability in signal recovery with downsampling or noncomplete data. The well-trained CNN can extract high-level abstract feature representation from the input data autonomously and exploit it in the signal recovery. We are interested in exploiting the CNN to enhance the CS ISAR imaging capability. The successful training of CNN always requires many thousand annotated training samples. This limits the application of CNN to the radar imaging field where large amount of training data cannot be obtained as easy as in other fields, e.g., computer vision. We propose a fully CNN (FCNN) for ISAR imaging. The constructed FCNN has a multistage decomposition and multichannel filtering architecture and has no fully connected layers. It can work with very few training samples as compared to existing CNN-based imaging networks. The imaging results of real ISAR data show that the proposed FCNN-based ISAR imaging method outperforms the state-of-the-art CS ISAR imaging methods in both image quality and computational efficiency.
- 9 October 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Letters
- Vol. 17 (7) , 1203-1207
Inverse Synthetic Aperture Radar Imaging Using a Fully Convolutional Neural Network
The traditional inverse synthetic aperture radar (ISAR) imaging uses the range-Doppler (RD) type of methods. The compressive sensing (CS)-based ISAR imaging is capable of obtaining good target images of high contrast and less sidelobe with much less downsampling data. However, the real application of CS ISAR imaging is limited by the time-consuming iteration-based image reconstruction. The image quality is also limited by the performance of sparse representation of the target scene. In recent years, deep learning methods, more specifically the convolutional neural network (CNN), has shown its capability in signal recovery with downsampling or noncomplete data. The well-trained CNN can extract high-level abstract feature representation from the input data autonomously and exploit it in the signal recovery. We are interested in exploiting the CNN to enhance the CS ISAR imaging capability. The successful training of CNN always requires many thousand annotated training samples. This limits the application of CNN to the radar imaging field where large amount of training data cannot be obtained as easy as in other fields, e.g., computer vision. We propose a fully CNN (FCNN) for ISAR imaging. The constructed FCNN has a multistage decomposition and multichannel filtering architecture and has no fully connected layers. It can work with very few training samples as compared to existing CNN-based imaging networks. The imaging results of real ISAR data show that the proposed FCNN-based ISAR imaging method outperforms the state-of-the-art CS ISAR imaging methods in both image quality and computational efficiency.
- 17 May 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Letters
- Vol. 16 (12) , 1849-1853
Subspace Learning Network: An Efficient ConvNet for PolSAR Image Classification
Land cover classification is an important part of the polarimetric synthetic aperture radar (PolSAR) image interpretation. The convolutional neural network (CNN) has been utilized to improve the classification accuracy recently. However, how to efficiently train the classification model with limited training samples while keeping the generalization performance is still a challenge. In this letter, we devise a subspace learning network (SSLNet) for PolSAR image classification, which can be trained more efficiently. First, a third-order polarimetric feature tensor is constructed using five-target decompositions to make full use of the prior knowledge. The tensor is then fed into a two-layer CNN in which the principal component analysis (PCA) is employed to learn the convolutional filters. Finally, the output features of the network are obtained by binary hashing and block-wise histograms, followed by the nearest neighbor (NN) classifier to complete the classification. Due to the simple learning strategy, the proposed SSLNet can be easily designed and efficiently trained. Experimental results on benchmark PolSAR data reveal that the SSLNet can achieve higher classification accuracy with limited training samples than the conventional CNN method.
- 17 May 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Letters
- Vol. 16 (12) , 1849-1853
Subspace Learning Network: An Efficient ConvNet for PolSAR Image Classification
Land cover classification is an important part of the polarimetric synthetic aperture radar (PolSAR) image interpretation. The convolutional neural network (CNN) has been utilized to improve the classification accuracy recently. However, how to efficiently train the classification model with limited training samples while keeping the generalization performance is still a challenge. In this letter, we devise a subspace learning network (SSLNet) for PolSAR image classification, which can be trained more efficiently. First, a third-order polarimetric feature tensor is constructed using five-target decompositions to make full use of the prior knowledge. The tensor is then fed into a two-layer CNN in which the principal component analysis (PCA) is employed to learn the convolutional filters. Finally, the output features of the network are obtained by binary hashing and block-wise histograms, followed by the nearest neighbor (NN) classifier to complete the classification. Due to the simple learning strategy, the proposed SSLNet can be easily designed and efficiently trained. Experimental results on benchmark PolSAR data reveal that the SSLNet can achieve higher classification accuracy with limited training samples than the conventional CNN method.
- 1 October 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-5
Inverse Synthetic Aperture Radar Imaging Using Group Based Dictionary Learning
As compared with traditional radar imaging methods, the compressive sensing (CS) radar imaging methods can obtain high-quality images using much less data. However, the default assumption that the target scene is sparse actually limits the performance of the CS based ISAR imaging methods. The dictionary learning (DL) has been used to find the sparse representation of the target scene where a block processing is used, but each image patch is considered independently and the relationship between the patches is not utilized. In this paper, we propose an improved DL method for ISAR imaging exploiting the idea of group sparsity. First, we use image patches with similar structure to construct several groups. Then, we utilize the SVD technique to learn a sparse transform inferred from the image patch groups. This learnt sparse transform is used to sparsely represent the target scene and the image reconstruction is performed. The experimental results show that the proposed group DL based ISAR imaging method can provide better imaging results of the target scene than the existing DL based CS ISAR reconstruction algorithms with higher computational efficiency.
- 1 October 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-5
Inverse Synthetic Aperture Radar Imaging Using Group Based Dictionary Learning
As compared with traditional radar imaging methods, the compressive sensing (CS) radar imaging methods can obtain high-quality images using much less data. However, the default assumption that the target scene is sparse actually limits the performance of the CS based ISAR imaging methods. The dictionary learning (DL) has been used to find the sparse representation of the target scene where a block processing is used, but each image patch is considered independently and the relationship between the patches is not utilized. In this paper, we propose an improved DL method for ISAR imaging exploiting the idea of group sparsity. First, we use image patches with similar structure to construct several groups. Then, we utilize the SVD technique to learn a sparse transform inferred from the image patch groups. This learnt sparse transform is used to sparsely represent the target scene and the image reconstruction is performed. The experimental results show that the proposed group DL based ISAR imaging method can provide better imaging results of the target scene than the existing DL based CS ISAR reconstruction algorithms with higher computational efficiency.
- 22 February 2018
- Published by SPIE-Intl Soc Optical Eng
- Vol. 10548, 5
Controllable low-loss slow light in photonic crystals
The key figures of merit for integrated optical components include the device footprint and operating energy consumption, both being major contributors to the purchase and operating cost, respectively. Slow light in silicon photonic crystal (PhC) devices has the potential to significantly reduce both, through enhanced light matter interactions, for both linear and non-linear optics applications. However, for all applications a precise control over the slow-down factor and reduced optical losses are paramount. In this paper, we present our work on various low-loss slow light systems based on PhC technology. We discuss the ability to control the group index and propagation loss of a PhC waveguide through appropriate device design - dispersion and loss engineered waveguides. This control, providing us with a free choice of group index ranging from 5 to 100, has already led to a range of non-linear optical applications, such as third harmonic generation, four-wave mixing and photon pair generation. We extend this approach to kagome lattice based PhCs and show that group indices exceeding 100 000 are possible in photonic crystal based geometries. We further discuss the post-fabrication control over slow light in PhC waveguides. Here both permanent, passive control is possible - through post-processing of the PhC devices - and adaptable, active control, through electro-optic or thermo-optic tuning. We apply the latter to a coupled cavity geometry that displays a transmission peak analogous to electromagnetically induced transparency and show a tuneable delay of 300ps with a delay loss of approximately 15dB/ns.
- 22 February 2018
- Published by SPIE-Intl Soc Optical Eng
- Vol. 10548, 5
Controllable low-loss slow light in photonic crystals
The key figures of merit for integrated optical components include the device footprint and operating energy consumption, both being major contributors to the purchase and operating cost, respectively. Slow light in silicon photonic crystal (PhC) devices has the potential to significantly reduce both, through enhanced light matter interactions, for both linear and non-linear optics applications. However, for all applications a precise control over the slow-down factor and reduced optical losses are paramount. In this paper, we present our work on various low-loss slow light systems based on PhC technology. We discuss the ability to control the group index and propagation loss of a PhC waveguide through appropriate device design - dispersion and loss engineered waveguides. This control, providing us with a free choice of group index ranging from 5 to 100, has already led to a range of non-linear optical applications, such as third harmonic generation, four-wave mixing and photon pair generation. We extend this approach to kagome lattice based PhCs and show that group indices exceeding 100 000 are possible in photonic crystal based geometries. We further discuss the post-fabrication control over slow light in PhC waveguides. Here both permanent, passive control is possible - through post-processing of the PhC devices - and adaptable, active control, through electro-optic or thermo-optic tuning. We apply the latter to a coupled cavity geometry that displays a transmission peak analogous to electromagnetically induced transparency and show a tuneable delay of 300ps with a delay loss of approximately 15dB/ns.