Zhongxiao Lan
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Zhongxiao Lan has a total of 7 co-authors. Here are the top co-authors by number of shared publications:
- Mingju Chen
School of Automation and Information Engineering, Sichuan University of Science and Engineering, Yibin 644005, China
2 shared publications - Sihang Yi
School of Automation and Information Engineering, Sichuan University of Science & Engineering, Yibin 644002, China
2 shared publications - Zhengxu Duan
Yibin Research Institute of University of Electronic Science and Technology of China, Yibin 644002, People’s Republic of China
2 shared publications Mingju Chen
Artificial Intelligence Key Laboratory of Sichuan Province, Sichuan University of Science & Engineering, Yibin 644002, China
1 shared publicationQin Su
School of Automation and Information Engineering, Sichuan University of Science & Engineering, Yibin 644002, China
1 shared publicationSihang Yi
School of Automation and Information Engineering, Sichuan University of Science & Engineering, Yibin 644002, China
1 shared publication
Latest Publications
- Open Access
An Efficient Image Deblurring Network with a Hybrid Architecture
Blurring is one of the main degradation factors in image degradation, so image deblurring is of great interest as a fundamental problem in low-level computer vision. Because of the limited receptive field, traditional CNNs lack global fuzzy region modeling, and do not make full use of rich context information between features. Recently, a transformer-based neural network structure has performed well in natural language tasks, inspiring rapid development in the field of defuzzification. Therefore, in this paper, a hybrid architecture based on CNN and transformers is used for image deblurring. Specifically, we first extract the shallow features of the blurred images using a cross-layer feature fusion block that emphasizes the contextual information of each feature extraction layer. Secondly, an efficient transformer module for extracting deep features is designed, which fully aggregates feature information at medium and long distances using vertical and horizontal intra- and inter-strip attention layers, and a dual gating mechanism is used as a feedforward neural network, which effectively reduces redundant features. Finally, the cross-layer feature fusion block is used to complement the feature information to obtain the deblurred image. Extensive experimental results on publicly available benchmark datasets GoPro, HIDE, and the real dataset RealBlur show that the proposed method outperforms the current mainstream deblurring algorithms and recovers the edge contours and texture details of the images more clearly.
Open AccessAn Efficient Image Deblurring Network with a Hybrid Architecture
Blurring is one of the main degradation factors in image degradation, so image deblurring is of great interest as a fundamental problem in low-level computer vision. Because of the limited receptive field, traditional CNNs lack global fuzzy region modeling, and do not make full use of rich context information between features. Recently, a transformer-based neural network structure has performed well in natural language tasks, inspiring rapid development in the field of defuzzification. Therefore, in this paper, a hybrid architecture based on CNN and transformers is used for image deblurring. Specifically, we first extract the shallow features of the blurred images using a cross-layer feature fusion block that emphasizes the contextual information of each feature extraction layer. Secondly, an efficient transformer module for extracting deep features is designed, which fully aggregates feature information at medium and long distances using vertical and horizontal intra- and inter-strip attention layers, and a dual gating mechanism is used as a feedforward neural network, which effectively reduces redundant features. Finally, the cross-layer feature fusion block is used to complement the feature information to obtain the deblurred image. Extensive experimental results on publicly available benchmark datasets GoPro, HIDE, and the real dataset RealBlur show that the proposed method outperforms the current mainstream deblurring algorithms and recovers the edge contours and texture details of the images more clearly.
- Open Access
- 24 May 2023
- Published by MDPI AG in Applied Sciences
- Vol. 13 (11) , 6407
Scene Reconstruction Algorithm for Unstructured Weak-Texture Regions Based on Stereo Vision
At present, Chinese 3D reconstruction solutions using stereo cameras mainly face known, indoor, structured scenes; for the reconstruction of unstructured, larger-scale scenes with a large variety of texture information of different intensities, there are certain difficulties in ensuring accuracy and real-time processing. For the above problems, we propose a scene reconstruction method using stereo vision. Firstly, considering the influence of outdoor lighting and weather on the captured 2D images, the optimized SAD-FAST feature detection algorithm and stereo-matching strategy were employed in the stereo-matching stage to improve the overall efficiency and matching quality at this stage. Then, a homogenized feature extraction algorithm with gradient value decreasing step by step (GVDS) was used in the depth value calculation to ensure a sufficient number of feature points for strong texture information while extracting features from weak-texture areas, which greatly improved the quality and speed of unstructured scene reconstruction. We conducted experiments to validate the proposed method, and the results showed the feasibility of the proposed method and its high practical value.
- 24 May 2023
- Published by MDPI AG in Applied Sciences
- Vol. 13 (11) , 6407
Open AccessScene Reconstruction Algorithm for Unstructured Weak-Texture Regions Based on Stereo Vision
At present, Chinese 3D reconstruction solutions using stereo cameras mainly face known, indoor, structured scenes; for the reconstruction of unstructured, larger-scale scenes with a large variety of texture information of different intensities, there are certain difficulties in ensuring accuracy and real-time processing. For the above problems, we propose a scene reconstruction method using stereo vision. Firstly, considering the influence of outdoor lighting and weather on the captured 2D images, the optimized SAD-FAST feature detection algorithm and stereo-matching strategy were employed in the stereo-matching stage to improve the overall efficiency and matching quality at this stage. Then, a homogenized feature extraction algorithm with gradient value decreasing step by step (GVDS) was used in the depth value calculation to ensure a sufficient number of feature points for strong texture information while extracting features from weak-texture areas, which greatly improved the quality and speed of unstructured scene reconstruction. We conducted experiments to validate the proposed method, and the results showed the feasibility of the proposed method and its high practical value.
- Open Access
- 1 January 2023
- Published by American Institute of Mathematical Sciences (AIMS) in Mathematical Biosciences and Engineering
- Vol. 20 (8) , 15476-15495
HDS-YOLOv5: An improved safety harness hook detection algorithm based on YOLOv5s
Improperly using safety harness hooks is a major factor of safety hazards during power maintenance operation. The machine vision-based traditional detection methods have low accuracy and limited real-time effectiveness. In order to quickly discern the status of hooks and reduce safety incidents in the complicated operation environments, three improvements are incorporated in YOLOv5s to construct the novel HDS-YOLOv5 network. First, HOOK-SPPF (spatial pyramid pooling fast) feature extraction module replaces the SPPF backbone network. It can enhance the network's feature extraction capability with less feature loss and extract more distinctive hook features from complex backgrounds. Second, a decoupled head module modified with confidence and regression frames is implemented to reduce negative conflicts between classification and regression, resulting in increased recognition accuracy and accelerated convergence. Lastly, the Scylla intersection over union (SIoU) is employed to optimize the loss function by utilizing the vector angle between the real and predicted frames, thereby improving the model's convergence. Experimental results demonstrate that the HDS-YOLOv5 algorithm achieves a 3% increase in mAP@0.5, reaching 91.2%. Additionally, the algorithm achieves a detection rate of 24.0 FPS (frames per second), demonstrating its superior performance compared to other models.
- 1 January 2023
- Published by American Institute of Mathematical Sciences (AIMS) in Mathematical Biosciences and Engineering
- Vol. 20 (8) , 15476-15495
Open AccessHDS-YOLOv5: An improved safety harness hook detection algorithm based on YOLOv5s
Improperly using safety harness hooks is a major factor of safety hazards during power maintenance operation. The machine vision-based traditional detection methods have low accuracy and limited real-time effectiveness. In order to quickly discern the status of hooks and reduce safety incidents in the complicated operation environments, three improvements are incorporated in YOLOv5s to construct the novel HDS-YOLOv5 network. First, HOOK-SPPF (spatial pyramid pooling fast) feature extraction module replaces the SPPF backbone network. It can enhance the network's feature extraction capability with less feature loss and extract more distinctive hook features from complex backgrounds. Second, a decoupled head module modified with confidence and regression frames is implemented to reduce negative conflicts between classification and regression, resulting in increased recognition accuracy and accelerated convergence. Lastly, the Scylla intersection over union (SIoU) is employed to optimize the loss function by utilizing the vector angle between the real and predicted frames, thereby improving the model's convergence. Experimental results demonstrate that the HDS-YOLOv5 algorithm achieves a 3% increase in mAP@0.5, reaching 91.2%. Additionally, the algorithm achieves a detection rate of 24.0 FPS (frames per second), demonstrating its superior performance compared to other models.