A5039949283- 3D Shape Modeling and Analysis
- Advanced Numerical Analysis Techniques
- 3D Surveying and Cultural Heritage
- Computer Graphics and Visualization Techniques
- Remote Sensing and LiDAR Applications
- Infrastructure Maintenance and Monitoring
- Asphalt Pavement Performance Evaluation
- Video Surveillance and Tracking Methods
- 3D Modeling in Geospatial Applications
- Face and Expression Recognition
- Human Pose and Action Recognition
- Concrete Corrosion and Durability
- Medical Imaging and Analysis
- Geotechnical Engineering and Underground Structures
- Face recognition and analysis
- Robotics and Sensor-Based Localization
Xihua University
2024
Xiamen University
2021-2023
Jimei University
2016
Mobile Laser Scanning (MLS) system can provide high-density and accurate 3D point clouds that enable rapid pavement crack detection for road maintenance tasks. Supervised learning-based algorithms have been proved pretty effective handling such a large amount of inhomogeneous unstructured clouds. However, these often rely on lot annotated data, which is labor-intensive time-consuming. This paper presents semi-supervised point-level approach to overcome this challenge. We propose graph-widen...
Accurate pavement crack detection is important for routine maintenance of pavements and reduction possible traffic accidents. Most existing rule- or learning-based point-level approaches cannot achieve high accuracy efficiency owing to the disorderly arrangement, scattered intensities, diverse structures, large data volumes, complex annotation mobile laser scanning (MLS) point clouds. To address these issues, we developed SCL-GCN, a Stratified Contrastive Learning Graph Convolution Network...
We propose a novel method called SHS-Net for oriented normal estimation of point clouds by learning signed hyper surfaces, which can accurately predict normals with global consistent orientation from various clouds. Almost all existing methods estimate through two-stage pipeline, i.e., unoriented and orientation, each step is implemented separate algorithm. However, previous are sensitive to parameter settings, resulting in poor results noise, density variations complex geometries. In this...
With the rapid development of Light Detection And Ranging (LiDAR) systems, novel dual-channel airborne LiDAR systems have emerged to provide more complete and precise data than traditional scanners for building instance extraction since 2013. RIEGL VQ-1560i, launched in 2016, is a state-of-the-art system, which capable capturing dense points on rooftops façades simultaneously, due unique innovative bidirectional scanning angle. Our proposed method first ever use subsequent point clouds...
We propose a novel method called SHS-Net for point cloud normal estimation by learning signed hyper surfaces, which can accurately predict normals with global consistent orientation from various clouds. Almost all existing methods estimate oriented through two-stage pipeline, i.e., unoriented and orientation, each step is implemented separate algorithm. However, previous are sensitive to parameter settings, resulting in poor results clouds noise, density variations complex geometries. In...
We propose Neural Gradient Learning (NGL), a deep learning approach to learn gradient vectors with consistent orientation from 3D point clouds for normal estimation. It has excellent approximation properties the underlying geometry of data. utilize simple neural network parameterize objective function produce gradients at points using global implicit representation. However, derived usually drift away ground-truth oriented normals due lack local detail descriptions. Therefore, we introduce...
This paper proposed an improved Local Binary Features (LBF) [1] algorithm for bayonet surveillance system. Since LBF is based on shape-regression strategy, which prone to over-fitting after multi-stage regression, the training model cannot be directly applied other scenarios. To this end, we employed new data into existing models at final stage of regression. As a consequent, newly imported can embedded model. The experimental results conducted videos showed that method outperforms LBF, ERT...
We propose a novel method called SHS-Net for oriented normal estimation of point clouds by learning signed hyper surfaces, which can accurately predict normals with global consistent orientation from various clouds. Almost all existing methods estimate through two-stage pipeline, i.e., unoriented and orientation, each step is implemented separate algorithm. However, previous are sensitive to parameter settings, resulting in poor results noise, density variations complex geometries. In this...
We propose Neural Gradient Learning (NGL), a deep learning approach to learn gradient vectors with consistent orientation from 3D point clouds for normal estimation. It has excellent approximation properties the underlying geometry of data. utilize simple neural network parameterize objective function produce gradients at points using global implicit representation. However, derived usually drift away ground-truth oriented normals due lack local detail descriptions. Therefore, we introduce...
Normal estimation for 3D point clouds is a fundamental task in geometry processing. The state-of-the-art methods rely on priors of fitting local surfaces learned from normal supervision. However, supervision benchmarks comes synthetic shapes and usually not available real scans, thereby limiting the these methods. In addition, orientation consistency across remains difficult to achieve without separate post-processing procedure. To resolve issues, we propose novel method estimating oriented...