Fusion of terrestrial and UAV-based point clouds for precision mapping and infrastructure inspection

目前，地面激光扫描仪（TLS）或基于无人机（UAV）的激光雷达/摄像系统用于精确测绘和基础设施检查。然而，不幸的是，这两种系统都有缺点，包括较低的空间分辨率和基于无人机的系统传感器集成的复杂性，以及移动性和无法扫描TLS的危险到达区域。**本研究项目旨在最佳地融合来自TLS和基于无人机的**激光雷达/相机系统的点云数据，这有可能克服单独使用任何系统的缺点。该研究将首先选择代表不同操作场景的不同地点（例如，桥梁，茂密的植被，建筑物），使用FARO TLS和基于无人机的系统进行数据收集。将建立多个地面控制点和目标，提高点云配准的图像采集和精度。由于将在无人机上使用低成本的数码相机，因此将开发一种准确有效的相机校准方法，用于所有可能的操作场景，包括确定内部定向元素和畸变参数。此外，激光雷达数据将根据几何和辐射效应进行校正。Trimble POSPac无人机软件包将用于** GNSS/MEMS IMU数据的后处理，以及成像和激光雷达数据的地理参考。Pix4D和Autodesk**ReCap Pro软件包将用于验证图像配准，以及MATLAB和**CloudCompare软件。TLS和基于无人机的点云数据最终将用于创建扫描结构和周围环境的完整3D模型。**本研究成果将显著提高现场工作流程效率，提高**定位精度，并加强加拿大基础设施的检查和风险评估程序。**这反过来又有助于在基础设施服务和维修方面做出明智的决定。

At present, either of the terrestrial laser scanners (TLS) or unmanned aerial vehicle (UAV)-based**LiDAR/camera systems are used for precision mapping and infrastructure inspection. Unfortunately, however,**both systems suffer from drawbacks, including the lower spatial resolution and complexity of sensor**integration of the UAV-based systems and the mobility and inability to scan dangerous-to-reach areas of TLS.**This research project aims to optimally fuse the point cloud data from both of the TLS and UAV-based**LiDAR/camera systems, which has the potential to overcome the drawbacks of either system alone.**The research will first select various sites representing different operational scenarios (e.g., bridges, dense**vegetation, buildings) for data collection with both of the FARO TLS and UAV-based systems. A number of**ground control points and targets will be established to enhance the image acquisition and accuracy of the point**cloud registration. As a low-cost digital camera will be used onboard the UAV, an accurate and efficient**camera calibration method for all possible operational scenarios will be developed, including the determination**interior orientation elements and distortion parameters. As well, LiDAR data will be corrected for both of the**geometric and radiometric effects. Trimble POSPac UAV software package will be used for post-processing of**the GNSS/MEMS IMU data and for geo-referencing of the imaging and LiDAR data. Pix4D and Autodesk**ReCap Pro software packages will be used to verify the image registration, along with MATLAB and**CloudCompare software. Both of the TLS and UAV-based point cloud data will finally be used to create a**complete 3D model of the scanned structure and surroundings.**The outcome of this research will significantly improve the efficiency of the field workflow, improve the**positioning accuracy, and enhance the inspection and risk assessment procedures of Canada's infrastructure.**This in turn helps making informed decision about infrastructure service and repair.