Xiankai Xue
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Xiankai Xue has a total of 7 co-authors. Here are the top co-authors by number of shared publications:
- Hao Li
CCCC Second Harbor Engineering Co., Ltd., Wuhan 430040, China
2 shared publications - Hao Zhu
CCCC Second Harbor Engineering Co., Ltd., Wuhan 430040, China
2 shared publications - Tao Jiang
CCCC Second Harbor Engineering Co., Ltd., Wuhan 430040, China
2 shared publications - Yakun Xie
Faculty of Geosciences and Environmental Engineering, Southwest Jiaotong University, Chengdu 610031, China
2 shared publications - Yongwei Wang
CCCC Second Harbor Engineering Co., Ltd., Wuhan 430040, China
2 shared publications - Mingqing Zhu
CCCC Second Harbor Engineering Co., Ltd., Wuhan 430040, China
1 shared publication
Latest Publications
- Open Access
- 13 November 2026
- Published by MDPI AG in Applied Sciences
- Vol. 16 (22) , 11250
Precise and Intelligent Registration of Multi-Station Point Clouds for Segmental Girders Based on Automatic Spherical Target Extraction and Two-Stage Radius Refinement
To address the issues of reliance on manual operation for spherical target extraction and limited sphere center positioning accuracy caused by the absence of radius priors in multi-station laser scanning point cloud registration for precast segmental girders, this paper proposes an intelligent multi-station point cloud registration method based on automatic spherical target extraction and two-stage radius refinement. Spherical targets are automatically extracted via surround-view projection and image cross-correlation, and point cloud backtracking is realized through pixel-to-3D point index mapping. Two-stage radius fitting is then conducted by integrating a prior radius knowledge base with robust statistical filtering: RANSAC and probabilistic filtering are first used to obtain the prior radius and feasible interval, which is subsequently treated as a hard constraint, and TRF and L-BFGS-B are then adopted for two-level optimization to refine the sphere centers. Finally, correspondences are established according to the spherical target distance matrix, and SVD is employed to solve the rigid transformation, thereby completing multi-station registration and fusion. Using a segmental girder for extraction, fitting, and comparative ablation experiments, the experimental results show that the proposed method is highly automated, with a 100% spherical target detection rate and an average matching degree higher than 0.80; the fitted radius differences range from 0 to 1.34 mm; the mean of the average consistency residuals of homologous sphere centers after registration is 1.26 mm; and the fitted radius difference and the corresponding segmental girder length difference are 0.52 mm and 3.5 mm, respectively, outperforming commonly used international fitting methods in accuracy. Three segmental girders were used for dimensional inspection validation, and the results show that the dimensional measurement errors of all three segmental girders meet the engineering tolerance requirements, verifying the generalization ability and stability of the proposed registration method.
- 13 November 2026
- Published by MDPI AG in Applied Sciences
- Vol. 16 (22) , 11250
Open AccessPrecise and Intelligent Registration of Multi-Station Point Clouds for Segmental Girders Based on Automatic Spherical Target Extraction and Two-Stage Radius Refinement
To address the issues of reliance on manual operation for spherical target extraction and limited sphere center positioning accuracy caused by the absence of radius priors in multi-station laser scanning point cloud registration for precast segmental girders, this paper proposes an intelligent multi-station point cloud registration method based on automatic spherical target extraction and two-stage radius refinement. Spherical targets are automatically extracted via surround-view projection and image cross-correlation, and point cloud backtracking is realized through pixel-to-3D point index mapping. Two-stage radius fitting is then conducted by integrating a prior radius knowledge base with robust statistical filtering: RANSAC and probabilistic filtering are first used to obtain the prior radius and feasible interval, which is subsequently treated as a hard constraint, and TRF and L-BFGS-B are then adopted for two-level optimization to refine the sphere centers. Finally, correspondences are established according to the spherical target distance matrix, and SVD is employed to solve the rigid transformation, thereby completing multi-station registration and fusion. Using a segmental girder for extraction, fitting, and comparative ablation experiments, the experimental results show that the proposed method is highly automated, with a 100% spherical target detection rate and an average matching degree higher than 0.80; the fitted radius differences range from 0 to 1.34 mm; the mean of the average consistency residuals of homologous sphere centers after registration is 1.26 mm; and the fitted radius difference and the corresponding segmental girder length difference are 0.52 mm and 3.5 mm, respectively, outperforming commonly used international fitting methods in accuracy. Three segmental girders were used for dimensional inspection validation, and the results show that the dimensional measurement errors of all three segmental girders meet the engineering tolerance requirements, verifying the generalization ability and stability of the proposed registration method.
- Open Access
PSDNet: A Bio-Inspired Perception-Based Method for Multi-Scale Semantic Segmentation of Non-Uniform Point Clouds of Precast Components
Three-dimensional laser-scanned point clouds of precast components are often affected by non-uniform density, large object-scale variation, and construction-site noise. Inspired by the dual-pathway perception of eagle vision, this study proposes PSDNet, an end-to-end semantic segmentation network for precast-component point clouds. An Eagle-Focused Density Intensification module (EFDI) enhances panel-edge details in sparsely sampled regions, while a Multi-Scale Global Fusion Module (MSGM) combines a one-time-built spatial index with global dual-pooling attention to distinguish objects ranging from centimeter-scale embedded components to meter-scale fixtures. Experiments were conducted on a dataset collected from a precast-component project in Riyadh, Saudi Arabia. PSDNet achieved a mean intersection-over-union of 73.6%, panel extraction accuracy of 90.9%, and embedded-component identification rate of 91.8%. Ablation experiments confirmed the complementary contributions of EFDI and MSGM. Comparisons with PointNet++, PointCNN, RandLA-Net, and KPConv showed consistently superior segmentation under non-uniform point densities and large-scale variations. The results demonstrate the potential of PSDNet for digital inspection of precast components.
Open AccessPSDNet: A Bio-Inspired Perception-Based Method for Multi-Scale Semantic Segmentation of Non-Uniform Point Clouds of Precast Components
Three-dimensional laser-scanned point clouds of precast components are often affected by non-uniform density, large object-scale variation, and construction-site noise. Inspired by the dual-pathway perception of eagle vision, this study proposes PSDNet, an end-to-end semantic segmentation network for precast-component point clouds. An Eagle-Focused Density Intensification module (EFDI) enhances panel-edge details in sparsely sampled regions, while a Multi-Scale Global Fusion Module (MSGM) combines a one-time-built spatial index with global dual-pooling attention to distinguish objects ranging from centimeter-scale embedded components to meter-scale fixtures. Experiments were conducted on a dataset collected from a precast-component project in Riyadh, Saudi Arabia. PSDNet achieved a mean intersection-over-union of 73.6%, panel extraction accuracy of 90.9%, and embedded-component identification rate of 91.8%. Ablation experiments confirmed the complementary contributions of EFDI and MSGM. Comparisons with PointNet++, PointCNN, RandLA-Net, and KPConv showed consistently superior segmentation under non-uniform point densities and large-scale variations. The results demonstrate the potential of PSDNet for digital inspection of precast components.