Zhijia Zhang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Zhijia Zhang has a total of 20 co-authors. Here are the top co-authors by number of shared publications:
- Dejian Li
School of Information Science and Engineering, Shenyang University of Technology, Shenyang, People’s Republic of China
2 shared publications - Guanghui Liu
School of Artificial Intelligence, Shenyang University of Technology, Shenyang, People’s Republic of China
2 shared publications - Jian Chen
School of Information Science and Engineering, Shenyang University of Technology, Shenyang, People’s Republic of China
2 shared publications Minmin Yang
The School of Artificial Intelligence, Shenyang University of Technology, Shenyang, China
2 shared publications- Minmin Yang
School of Artificial Intelligence, Shenyang University of Technology, Shenyang 110870, China
2 shared publications - Qiushuang Li
School of Artificial Intelligence, Shenyang University of Technology, Shenyang, People’s Republic of China
2 shared publications
Latest Publications
- 21 September 2026
- Published by Springer Nature in Multimedia Systems
- Vol. 32 (9) , 599
Dual-level modality mixing and spatial-temporal clues mining for video-based visible-infrared person re-identification
- 21 September 2026
- Published by Springer Nature in Multimedia Systems
- Vol. 32 (9) , 599
Dual-level modality mixing and spatial-temporal clues mining for video-based visible-infrared person re-identification
- Open Access
Correction: Shao et al. Symmetry-Guided AB-Dynamic Feature Refinement Network for Weakly Supervised Shadow Removal. Symmetry 2026, 18, 330
There was an error in the original publication
Open AccessCorrection: Shao et al. Symmetry-Guided AB-Dynamic Feature Refinement Network for Weakly Supervised Shadow Removal. Symmetry 2026, 18, 330
There was an error in the original publication
- 16 February 2026
- Published by Springer Nature in International Journal of Machine Learning and Cybernetics
- Vol. 17 (3) , 92
IGDBT: Illumination-guide dual branch transformer for low-light image enhancement
- 16 February 2026
- Published by Springer Nature in International Journal of Machine Learning and Cybernetics
- Vol. 17 (3) , 92
IGDBT: Illumination-guide dual branch transformer for low-light image enhancement
- Open Access
Symmetry-Guided AB-Dynamic Feature Refinement Network for Weakly Supervised Shadow Removal
Shadow removal aims to restore photometric, chromatic, and structural consistency between shadowed and non-shadowed image regions. Although weakly supervised shadow removal methods reduce the reliance on densely paired training data, they still struggle to fully exploit appearance priors from non-shadow regions. As a result, their shadow removal outputs often appear unnatural, exhibiting color shifts and loss of fine texture details. To address this issue, we propose an ab-dynamic feature refinement network (AB-DFRNet) for weakly supervised shadow removal that more effectively exploits structural and chromatic symmetry during training. A high-frequency information enhancement (HFIE) module is introduced into the shadow generation subnet to extract and enhance high-frequency components via frequency separation and dense convolutions, thereby facilitating the learning of fine structural symmetry and enriching pseudo-shadow details. In the removal subnet, a dual-attention adaptive fusion (DAAF) module combines global and local attention mechanisms to adaptively recalibrate channel-wise and spatial features, improving multi-scale feature integration. Furthermore, a chrominance-only consistency (COC) loss is designed to minimize differences between the a and b channels of restored regions and their non-shadow references in the Lab color space. This additional color refinement constraint encourages a symmetric distribution of chromatic information and helps the refinement network produce more natural shadow-removed results. Extensive experiments are conducted on three benchmark datasets: ISTD, SRD, and Video Shadow Removal. The results confirm the effectiveness of AB-DFRNet, demonstrating competitive quantitative performance and noticeably better visual quality compared with existing weakly supervised shadow removal methods.
Open AccessSymmetry-Guided AB-Dynamic Feature Refinement Network for Weakly Supervised Shadow Removal
Shadow removal aims to restore photometric, chromatic, and structural consistency between shadowed and non-shadowed image regions. Although weakly supervised shadow removal methods reduce the reliance on densely paired training data, they still struggle to fully exploit appearance priors from non-shadow regions. As a result, their shadow removal outputs often appear unnatural, exhibiting color shifts and loss of fine texture details. To address this issue, we propose an ab-dynamic feature refinement network (AB-DFRNet) for weakly supervised shadow removal that more effectively exploits structural and chromatic symmetry during training. A high-frequency information enhancement (HFIE) module is introduced into the shadow generation subnet to extract and enhance high-frequency components via frequency separation and dense convolutions, thereby facilitating the learning of fine structural symmetry and enriching pseudo-shadow details. In the removal subnet, a dual-attention adaptive fusion (DAAF) module combines global and local attention mechanisms to adaptively recalibrate channel-wise and spatial features, improving multi-scale feature integration. Furthermore, a chrominance-only consistency (COC) loss is designed to minimize differences between the a and b channels of restored regions and their non-shadow references in the Lab color space. This additional color refinement constraint encourages a symmetric distribution of chromatic information and helps the refinement network produce more natural shadow-removed results. Extensive experiments are conducted on three benchmark datasets: ISTD, SRD, and Video Shadow Removal. The results confirm the effectiveness of AB-DFRNet, demonstrating competitive quantitative performance and noticeably better visual quality compared with existing weakly supervised shadow removal methods.
- 4 December 2025
- Published by IOP Publishing in Measurement Science and Technology
- Vol. 36 (12) , 125403
Research on three dimensional reconstruction method based on noise loop detection and visual complementary fusion
Existing methods predominantly perform denoising and matching within a unified visual feature space, often neglecting multi-view positional and color information. This paper proposes a method based on noise loop detection and visual complementary fusion for high-precision three-dimensional (3D) reconstruction. Specifically, leveraging the association between the robotic arm coordinate system and the camera coordinate system, we first introduce a multi-level loop detection noise module. This module identifies and processes noise in the data in stages, mitigating its adverse effects on reconstruction accuracy. Subsequently, spatial transformations are applied to both RGB and depth images, establishing a robust initial foundation for subsequent processing. Finally, leveraging the complementary characteristics of both image types, we design an RGB and depth image collaborative registration (RDCR) module to maximize the strengths of each visual modality. Experiments on four public datasets and real-world components demonstrated a 1.26 mm reduction in contour reconstruction error, a 1.4% improvement in intersection over union, a −1.07 mm decrease in chamfer distance, and a 5.3% increase in completeness ratio compared to state-of-the-art methods. The proposed framework offers a novel approach for high-precision 3D visual reconstruction.
- 4 December 2025
- Published by IOP Publishing in Measurement Science and Technology
- Vol. 36 (12) , 125403
Research on three dimensional reconstruction method based on noise loop detection and visual complementary fusion
Existing methods predominantly perform denoising and matching within a unified visual feature space, often neglecting multi-view positional and color information. This paper proposes a method based on noise loop detection and visual complementary fusion for high-precision three-dimensional (3D) reconstruction. Specifically, leveraging the association between the robotic arm coordinate system and the camera coordinate system, we first introduce a multi-level loop detection noise module. This module identifies and processes noise in the data in stages, mitigating its adverse effects on reconstruction accuracy. Subsequently, spatial transformations are applied to both RGB and depth images, establishing a robust initial foundation for subsequent processing. Finally, leveraging the complementary characteristics of both image types, we design an RGB and depth image collaborative registration (RDCR) module to maximize the strengths of each visual modality. Experiments on four public datasets and real-world components demonstrated a 1.26 mm reduction in contour reconstruction error, a 1.4% improvement in intersection over union, a −1.07 mm decrease in chamfer distance, and a 5.3% increase in completeness ratio compared to state-of-the-art methods. The proposed framework offers a novel approach for high-precision 3D visual reconstruction.
- Open Access
- 4 November 2025
- Published by IOP Publishing in Engineering Research Express
- Vol. 7 (4) , 045253
Three-dimensional reconstruction of gearbox from multi-view point clouds with surface feature parfameter measurement method
Accurate reconstruction of the three-dimensional (3D) information of a gearbox and precise acquisition of its surface features are crucial for ensuring coating quality. This paper proposes a method for 3D reconstruction of gearbox surface and measurement of characteristic parameters based on multi-view point clouds. A multi-view, multi-rotation visual acquisition system based on the kinect sensor was built to obtain comprehensive point cloud data of the workpiece model from six views. Then, the acquired point cloud data was preprocessed. Normal vector filtering(NF), Fruit fly optimization algorithm(FOA) modified point cloud erosion method and fast point feature histogram(FPFH) algorithm were applied for point cloud coarse registration, and generalized iterative closest point(GICP) method for accurate registration, thus completing point cloud 3D reconstruction. Subsequently, the contour, part height and curvature feature parameters of the reconstructed model were measured. Compared with ground-truth values, the method yields mean absolute errors of 6.47 mm for workpiece contour, 1.056 mm for part height, and 0.107 m −1 for surface curvature, with corresponding relative accuracies of 98.7%, 93.125%, and 97.426%, respectively, where accuracy denotes the percentage of the computed mean relative to the actual mean. This study demonstrates the feasibility of the proposed approach for gearbox surface characterization and supports further 3D reconstruction and quantitative analysis. However, the current validation is limited to six gearboxes under controlled laboratory conditions; future work will address more challenging industrial scenarios involving oil contamination, occlusion, and variable reflections.
- 4 November 2025
- Published by IOP Publishing in Engineering Research Express
- Vol. 7 (4) , 045253
Open AccessThree-dimensional reconstruction of gearbox from multi-view point clouds with surface feature parfameter measurement method
Accurate reconstruction of the three-dimensional (3D) information of a gearbox and precise acquisition of its surface features are crucial for ensuring coating quality. This paper proposes a method for 3D reconstruction of gearbox surface and measurement of characteristic parameters based on multi-view point clouds. A multi-view, multi-rotation visual acquisition system based on the kinect sensor was built to obtain comprehensive point cloud data of the workpiece model from six views. Then, the acquired point cloud data was preprocessed. Normal vector filtering(NF), Fruit fly optimization algorithm(FOA) modified point cloud erosion method and fast point feature histogram(FPFH) algorithm were applied for point cloud coarse registration, and generalized iterative closest point(GICP) method for accurate registration, thus completing point cloud 3D reconstruction. Subsequently, the contour, part height and curvature feature parameters of the reconstructed model were measured. Compared with ground-truth values, the method yields mean absolute errors of 6.47 mm for workpiece contour, 1.056 mm for part height, and 0.107 m −1 for surface curvature, with corresponding relative accuracies of 98.7%, 93.125%, and 97.426%, respectively, where accuracy denotes the percentage of the computed mean relative to the actual mean. This study demonstrates the feasibility of the proposed approach for gearbox surface characterization and supports further 3D reconstruction and quantitative analysis. However, the current validation is limited to six gearboxes under controlled laboratory conditions; future work will address more challenging industrial scenarios involving oil contamination, occlusion, and variable reflections.
- 19 October 2025
- Published by Elsevier in Digital Signal Processing
- Vol. 168
Unsupervised visible-infrared person re-identification with pseudo-label refinement and modality-unified memory
- 19 October 2025
- Published by Elsevier in Digital Signal Processing
- Vol. 168
Unsupervised visible-infrared person re-identification with pseudo-label refinement and modality-unified memory
- Open Access
A Fast and Accurate Lane Detection Method Based on Row Anchor and Transformer Structure
Lane detection plays a pivotal role in the successful implementation of Advanced Driver Assistance Systems (ADASs), which are essential for detecting the road’s lane markings and determining the vehicle’s position, thereby influencing subsequent decision making. However, current deep learning-based lane detection methods encounter challenges. Firstly, the on-board hardware limitations necessitate an exceptionally fast prediction speed for the lane detection method. Secondly, improvements are required for effective lane detection in complex scenarios. This paper addresses these issues by enhancing the row-anchor-based lane detection method. The Transformer encoder–decoder structure is leveraged as the row classification enhances the model’s capability to extract global features and detect lane lines in intricate environments. The Feature-aligned Pyramid Network (FaPN) structure serves as an auxiliary branch, complemented by a novel structural loss with expectation loss, further refining the method’s accuracy. The experimental results demonstrate our method’s commendable accuracy and real-time performance, achieving a rapid prediction speed of 129 FPS (the single prediction time of the model on RTX3080 is 15.72 ms) and a 96.16% accuracy on the Tusimple dataset—a 3.32% improvement compared to the baseline method.
Open AccessA Fast and Accurate Lane Detection Method Based on Row Anchor and Transformer Structure
Lane detection plays a pivotal role in the successful implementation of Advanced Driver Assistance Systems (ADASs), which are essential for detecting the road’s lane markings and determining the vehicle’s position, thereby influencing subsequent decision making. However, current deep learning-based lane detection methods encounter challenges. Firstly, the on-board hardware limitations necessitate an exceptionally fast prediction speed for the lane detection method. Secondly, improvements are required for effective lane detection in complex scenarios. This paper addresses these issues by enhancing the row-anchor-based lane detection method. The Transformer encoder–decoder structure is leveraged as the row classification enhances the model’s capability to extract global features and detect lane lines in intricate environments. The Feature-aligned Pyramid Network (FaPN) structure serves as an auxiliary branch, complemented by a novel structural loss with expectation loss, further refining the method’s accuracy. The experimental results demonstrate our method’s commendable accuracy and real-time performance, achieving a rapid prediction speed of 129 FPS (the single prediction time of the model on RTX3080 is 15.72 ms) and a 96.16% accuracy on the Tusimple dataset—a 3.32% improvement compared to the baseline method.