Changjun Zou
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
Changjun Zou has a total of 19 co-authors. Here are the top co-authors by number of shared publications:
Lintao Ye
Affiliation East China Jiaotong University, Nanchang, China
6 shared publicationsHangbin Xu
School of Information and Software Engineering, East China Jiaotong University, Nanchang, China
3 shared publications- ChuChao Lin
School of Information and Software Engineering East China Jiaotong University Jiangxi China
2 shared publications Chuchao Lin
School of Information and Software Engineering, East China Jiaotong University, Nanchang, China
2 shared publications- HangBin Xu
School of Information and Software Engineering East China Jiaotong University Jiangxi China
2 shared publications - Jia Yu
School of Information Engineering, East China Jiaotong University, Nanchang 330013, China
2 shared publications
Latest Publications
- Open Access
Improved WCSPH-DEM Coupling for Analyzing Fluid–Solid Interactions
Fluid–structure interaction (FSI) research is crucial for applications in fields such as naval engineering, geological hazards, and biomechanics. Traditional grid-based methods (such as CFD) often face challenges in simulating large-deformation flow fields and complex boundary conditions, where mesh distortion can compromise simulation accuracy. Building upon the DualSPHysics5.2 framework, this study leverages the strengths of weakly compressible SPH (WCSPH) in modeling free surface flows and large-deformation fluids, as well as the discrete element method (DEM), for accurately describing particle collisions and fragmentation behaviors. We propose an improved MSPH-DEM coupling algorithm that incorporates moving least squares (MLS) correction for kernel function gradient optimization. This algorithm utilizes MLS-based gradient correction to achieve smoother fluid surfaces as well as bidirectional coupling between fluids and particles. Experimental validation demonstrates that in dam break simulations, this method reduces pressure errors. In the dam break impacting a cube experiment, it enhances accuracy, while in the dam break impacting a baffle experiment, the horizontal displacement of marker points closely aligns with the experimental values from Liao et al. This approach effectively improves the accuracy of the simulations of FSI problems, offering a more reliable numerical simulation methodology for engineering applications such as geological hazard prevention.
Open AccessImproved WCSPH-DEM Coupling for Analyzing Fluid–Solid Interactions
Fluid–structure interaction (FSI) research is crucial for applications in fields such as naval engineering, geological hazards, and biomechanics. Traditional grid-based methods (such as CFD) often face challenges in simulating large-deformation flow fields and complex boundary conditions, where mesh distortion can compromise simulation accuracy. Building upon the DualSPHysics5.2 framework, this study leverages the strengths of weakly compressible SPH (WCSPH) in modeling free surface flows and large-deformation fluids, as well as the discrete element method (DEM), for accurately describing particle collisions and fragmentation behaviors. We propose an improved MSPH-DEM coupling algorithm that incorporates moving least squares (MLS) correction for kernel function gradient optimization. This algorithm utilizes MLS-based gradient correction to achieve smoother fluid surfaces as well as bidirectional coupling between fluids and particles. Experimental validation demonstrates that in dam break simulations, this method reduces pressure errors. In the dam break impacting a cube experiment, it enhances accuracy, while in the dam break impacting a baffle experiment, the horizontal displacement of marker points closely aligns with the experimental values from Liao et al. This approach effectively improves the accuracy of the simulations of FSI problems, offering a more reliable numerical simulation methodology for engineering applications such as geological hazard prevention.
- Open Access
FLIP-IBM: Fluid–Structure Coupling Interaction Based on Immersed Boundary Method Under FLIP Framework
Fluid–structure coupling is a prominent and hot topic in computer graphics and virtual reality. The hybrid technique known as FLIP combines the benefits of grid-based and particle-based techniques. Nevertheless, a significant problem is figuring out how to accomplish fluid–structure coupling interaction based on the FLIP technique framework. We propose an immersed boundary approach to handle the problem of realistic fluid–structure coupling interaction under the FLIP framework. The benchmark test results demonstrate that, in addition to producing rich fluid–structure coupling interaction results, our novel technique also effectively reflects the effects of moving obstacle boundaries on the flow and pressure fields, thereby expanding the application area of the FLIP method.
Open AccessFLIP-IBM: Fluid–Structure Coupling Interaction Based on Immersed Boundary Method Under FLIP Framework
Fluid–structure coupling is a prominent and hot topic in computer graphics and virtual reality. The hybrid technique known as FLIP combines the benefits of grid-based and particle-based techniques. Nevertheless, a significant problem is figuring out how to accomplish fluid–structure coupling interaction based on the FLIP technique framework. We propose an immersed boundary approach to handle the problem of realistic fluid–structure coupling interaction under the FLIP framework. The benchmark test results demonstrate that, in addition to producing rich fluid–structure coupling interaction results, our novel technique also effectively reflects the effects of moving obstacle boundaries on the flow and pressure fields, thereby expanding the application area of the FLIP method.
- 26 May 2025
- Published by Wiley in Computer Animation and Virtual Worlds
- Vol. 36 (3)
UTMCR: 3U‐Net Transformer With Multi‐Contrastive Regularization for Single Image Dehazing
Convolutional neural networks have a long history of development in single‐width dehazing tasks, but have gradually been dominated by the Transformer framework due to their insufficient global modeling capability and large number of parameters. However, the existing Transformer network structure adopts a single U‐Net structure, which is insufficient in multi‐level and multi‐scale feature fusion and modeling capability. Therefore, we propose an end‐to‐end dehazing network (UTMCR‐Net). The network consists of two parts: (1) UT module, which connects three U‐Net networks in series, where the backbone is replaced by the Dehazeformer block. By connecting three U‐Net networks in series, we can improve the image global modeling capability and capture multi‐scale information at different levels to achieve multi‐level and multi‐scale feature fusion. (2) MCR module, which improves the original contrastive regularization method by splitting the results of the UT module into four equal blocks, which are then compared and learned by using the contrast regularization method, respectively. Specifically, we use three U‐Net networks to enhance the global modeling capability of UTMCR as well as the multi‐scale feature fusion capability. The image dehazing ability is further enhanced using the MCR module. Experimental results show that our method achieves better results on most datasets.
- 26 May 2025
- Published by Wiley in Computer Animation and Virtual Worlds
- Vol. 36 (3)
UTMCR: 3U‐Net Transformer With Multi‐Contrastive Regularization for Single Image Dehazing
Convolutional neural networks have a long history of development in single‐width dehazing tasks, but have gradually been dominated by the Transformer framework due to their insufficient global modeling capability and large number of parameters. However, the existing Transformer network structure adopts a single U‐Net structure, which is insufficient in multi‐level and multi‐scale feature fusion and modeling capability. Therefore, we propose an end‐to‐end dehazing network (UTMCR‐Net). The network consists of two parts: (1) UT module, which connects three U‐Net networks in series, where the backbone is replaced by the Dehazeformer block. By connecting three U‐Net networks in series, we can improve the image global modeling capability and capture multi‐scale information at different levels to achieve multi‐level and multi‐scale feature fusion. (2) MCR module, which improves the original contrastive regularization method by splitting the results of the UT module into four equal blocks, which are then compared and learned by using the contrast regularization method, respectively. Specifically, we use three U‐Net networks to enhance the global modeling capability of UTMCR as well as the multi‐scale feature fusion capability. The image dehazing ability is further enhanced using the MCR module. Experimental results show that our method achieves better results on most datasets.
- Top Cited Papers
- 1 May 2025
- Published by Wiley in Computer Animation and Virtual Worlds
- Vol. 36 (3)
SCNet: A Dual‐Branch Network for Strong Noisy Image Denoising Based on Swin Transformer and ConvNeXt
Image denoising plays a vital role in restoring high‐quality images from noisy inputs and directly impacts downstream vision tasks. Traditional methods often fail under strong noise, causing detail loss or excessive smoothing. While recent Convolutional Neural Networks‐based and Transformer‐based models have shown progress, they struggle to jointly capture global structure and preserve local details. To address this, we propose SCNet, a dual‐branch fusion network tailored for strong‐noise denoising. It combines a Swin Transformer branch for global context modeling and a ConvNeXt branch for fine‐grained local feature extraction. Their outputs are adaptively merged via a Feature Fusion Block using joint spatial and channel attention, ensuring semantic consistency and texture fidelity. A multi‐scale upsampling module and the Charbonnier loss further improve structural accuracy and visual quality. Extensive experiments on four benchmark datasets show that SCNet outperforms state‐of‐the‐art methods, especially under severe noise, and proves effective in real‐world tasks such as mural image restoration.
- 1 May 2025
- Published by Wiley in Computer Animation and Virtual Worlds
- Vol. 36 (3)
Top Cited PapersSCNet: A Dual‐Branch Network for Strong Noisy Image Denoising Based on Swin Transformer and ConvNeXt
Image denoising plays a vital role in restoring high‐quality images from noisy inputs and directly impacts downstream vision tasks. Traditional methods often fail under strong noise, causing detail loss or excessive smoothing. While recent Convolutional Neural Networks‐based and Transformer‐based models have shown progress, they struggle to jointly capture global structure and preserve local details. To address this, we propose SCNet, a dual‐branch fusion network tailored for strong‐noise denoising. It combines a Swin Transformer branch for global context modeling and a ConvNeXt branch for fine‐grained local feature extraction. Their outputs are adaptively merged via a Feature Fusion Block using joint spatial and channel attention, ensuring semantic consistency and texture fidelity. A multi‐scale upsampling module and the Charbonnier loss further improve structural accuracy and visual quality. Extensive experiments on four benchmark datasets show that SCNet outperforms state‐of‐the‐art methods, especially under severe noise, and proves effective in real‐world tasks such as mural image restoration.
- Open Access
- 7 March 2025
- Published by IOS Press
An Improved Fluid-Structure Coupling Interaction Method Based on SPH Method
This article proposes an improved method, aiming to significantly reduce computational time and reduce eddy current dissipation. This method has shown significant advantages in dam break, liquid sloshing, and floating wave experiments, especially in floating wave experiments where it is almost identical to real data. In addition, the DBC boundary method has been improved by adjusting the particle distribution to improve the accuracy, stability, and efficiency of the simulation. In the floating wave experiment, the author also used passive absorption wave technology to improve measurement accuracy, and based on this method, measured the pressure in the liquid sloshing experiment. This method more accurately captures the details of fluid flow and describes complex boundary movements through dynamic boundary methods, thereby improving the efficiency and accuracy of simulation. Compared to traditional SPH methods, this method avoids serious numerical dissipation problems. thereby reducing the risk of simulation failure. the experiments significantly contribute to the study of flow characteristics and the formation of shock waves. They also enable better design and maintenance of hydraulic structures, dams, and flood control facilities, thereby enhancing their safety and reliability.
- 7 March 2025
- Published by IOS Press
Open AccessAn Improved Fluid-Structure Coupling Interaction Method Based on SPH Method
This article proposes an improved method, aiming to significantly reduce computational time and reduce eddy current dissipation. This method has shown significant advantages in dam break, liquid sloshing, and floating wave experiments, especially in floating wave experiments where it is almost identical to real data. In addition, the DBC boundary method has been improved by adjusting the particle distribution to improve the accuracy, stability, and efficiency of the simulation. In the floating wave experiment, the author also used passive absorption wave technology to improve measurement accuracy, and based on this method, measured the pressure in the liquid sloshing experiment. This method more accurately captures the details of fluid flow and describes complex boundary movements through dynamic boundary methods, thereby improving the efficiency and accuracy of simulation. Compared to traditional SPH methods, this method avoids serious numerical dissipation problems. thereby reducing the risk of simulation failure. the experiments significantly contribute to the study of flow characteristics and the formation of shock waves. They also enable better design and maintenance of hydraulic structures, dams, and flood control facilities, thereby enhancing their safety and reliability.
- 12 December 2024
- Published by Springer Nature in Journal of Shanghai Jiaotong University (Science)
- Vol. 31 (4) , 1-13
Swin2FII: Fluid Image Inpainting Method Based on SwinV2 Transformer and Fast Fourier Convolution
- 12 December 2024
- Published by Springer Nature in Journal of Shanghai Jiaotong University (Science)
- Vol. 31 (4) , 1-13
Swin2FII: Fluid Image Inpainting Method Based on SwinV2 Transformer and Fast Fourier Convolution
- Open Access
- 14 August 2023
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 18 (8) , e0286711
Multi-channel feature fusion attention Dehazing network
Haze is a typical weather phenomena that has a significant negative impact on transportation safety, particularly in the port, highways, and airport runway areas. A multi-scale U-shaped dehazing network is proposed in this research, which is based on our multi-channel feature fusion attention structure. With the help of the feature fusion attention techniques, the model can focus on the intriguing locations with higher haze concentration area. In conjunction with UNet, it can achieve multi-scale feature reuse and residual learning, allowing it to fully utilize the feature information of each layer for image restoration. Experimental resulsts show that our technique performs well on a variety of test datasets. On highway data sets, the PSNR / SSIM / L∞ error performance over the novel technique is increased by 0.52% / 0.5% / 30.84%, 4.68% / 0.78% / 26.19% and 13.84% / 9.05% / 55.57% respectively, when compared to DehazeFormer, MIRNetv2, and FSDGN methods. The findings suggest that our proposed method performs better on image dehazing, especially in terms of L∞ error performance.
- 14 August 2023
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 18 (8) , e0286711
Open AccessMulti-channel feature fusion attention Dehazing network
Haze is a typical weather phenomena that has a significant negative impact on transportation safety, particularly in the port, highways, and airport runway areas. A multi-scale U-shaped dehazing network is proposed in this research, which is based on our multi-channel feature fusion attention structure. With the help of the feature fusion attention techniques, the model can focus on the intriguing locations with higher haze concentration area. In conjunction with UNet, it can achieve multi-scale feature reuse and residual learning, allowing it to fully utilize the feature information of each layer for image restoration. Experimental resulsts show that our technique performs well on a variety of test datasets. On highway data sets, the PSNR / SSIM / L∞ error performance over the novel technique is increased by 0.52% / 0.5% / 30.84%, 4.68% / 0.78% / 26.19% and 13.84% / 9.05% / 55.57% respectively, when compared to DehazeFormer, MIRNetv2, and FSDGN methods. The findings suggest that our proposed method performs better on image dehazing, especially in terms of L∞ error performance.
- 27 July 2023
- Published by Springer Nature
- p. 584-594
A Survey of Image Super-Resolution Reconstruction Based on Deep Learning
- 27 July 2023
- Published by Springer Nature
- p. 584-594
A Survey of Image Super-Resolution Reconstruction Based on Deep Learning
- Open Access
- 18 January 2023
- Published by Springer Nature in Research Square
Image Super-Resolution with Multi-scale Residual Dense Feature aggregation Networks
In recent years, for the effect of image reconstruction, the research of single image Super-resolution (SISR) using Convolutional Neural Networks (CNN) has made significant progress, and is superior to traditional methods. Aiming at improving network performance, residual connection plays a key role in these CNN-based methods. With the increase of the depth of these CNN-based networks, the residual features more aggregate the information of multiple different levels of the input image, which is more advantageous for the reconstruction of image details. However, inmost deep CNN-based SR models, the hierarchical features of the original Low-resolution (LR) image are not fully utilized. In order to solve this problem. We propose Residual Dense Feature Aggregation Block (RDFAB). Where RDFAB is the unit of residual blocks and the features extracted from each residual block are aggregated to the end of the module. As a result, RDFAB is able to produce more representative features. To further strengthen the effectiveness of RDFAB, we propose a multi-scale feature aggregation (MFA) framework, and our final multi-scale residual dense feature aggregation network (MRDFAN) combines RDFAB and MFa together with the addition of skip connections.Comprehensive experiments demonstrate the necessity of the MF A framework and the advantages of MRDFAN over the state of the art SISR methods.
- 18 January 2023
- Published by Springer Nature in Research Square
Open AccessImage Super-Resolution with Multi-scale Residual Dense Feature aggregation Networks
In recent years, for the effect of image reconstruction, the research of single image Super-resolution (SISR) using Convolutional Neural Networks (CNN) has made significant progress, and is superior to traditional methods. Aiming at improving network performance, residual connection plays a key role in these CNN-based methods. With the increase of the depth of these CNN-based networks, the residual features more aggregate the information of multiple different levels of the input image, which is more advantageous for the reconstruction of image details. However, inmost deep CNN-based SR models, the hierarchical features of the original Low-resolution (LR) image are not fully utilized. In order to solve this problem. We propose Residual Dense Feature Aggregation Block (RDFAB). Where RDFAB is the unit of residual blocks and the features extracted from each residual block are aggregated to the end of the module. As a result, RDFAB is able to produce more representative features. To further strengthen the effectiveness of RDFAB, we propose a multi-scale feature aggregation (MFA) framework, and our final multi-scale residual dense feature aggregation network (MRDFAN) combines RDFAB and MFa together with the addition of skip connections.Comprehensive experiments demonstrate the necessity of the MF A framework and the advantages of MRDFAN over the state of the art SISR methods.
- Open Access
- 31 December 2022
- Published by Wiley in Mobile Information Systems
- Vol. 2022, 1-11
Multichannel Cross-Scale Semantic Coherent Attention Network for Image Inpainting
This paper investigates a cross-scale space semantic feature coherent image inpainting approach since it is challenging for the existing image inpainting methods to fuse the semantic feature information effectively. Firstly, we learn the feature semantic relevance step-by-step from the high-level semantic feature map’s attention mechanism and then we apply what we have learned to the preceding low-level feature map. In order to preserve the visual and semantic coherence of image repair, the missing content can be filled by changing attention from deep to shallow in a multiscale manner. A broader receptive field is generated by partial convolution, and semantic feature relevance is achieved using a multiscale cross feature space feature attention mechanism based on semantic attention. This technique improves the extensibility and continuity of the restored images by reconstructing the semantic information of different feature spaces, not only taking into account the reuse of existing semantic space features but also including across feature spaces. The experimental results demonstrated an improvement in PSNR, SSIM, and L1 performance by 10.50%, 0.13%, and 47.09%, respectively, with clear benefits.
- 31 December 2022
- Published by Wiley in Mobile Information Systems
- Vol. 2022, 1-11
Open AccessMultichannel Cross-Scale Semantic Coherent Attention Network for Image Inpainting
This paper investigates a cross-scale space semantic feature coherent image inpainting approach since it is challenging for the existing image inpainting methods to fuse the semantic feature information effectively. Firstly, we learn the feature semantic relevance step-by-step from the high-level semantic feature map’s attention mechanism and then we apply what we have learned to the preceding low-level feature map. In order to preserve the visual and semantic coherence of image repair, the missing content can be filled by changing attention from deep to shallow in a multiscale manner. A broader receptive field is generated by partial convolution, and semantic feature relevance is achieved using a multiscale cross feature space feature attention mechanism based on semantic attention. This technique improves the extensibility and continuity of the restored images by reconstructing the semantic information of different feature spaces, not only taking into account the reuse of existing semantic space features but also including across feature spaces. The experimental results demonstrated an improvement in PSNR, SSIM, and L1 performance by 10.50%, 0.13%, and 47.09%, respectively, with clear benefits.