基于移动式激光扫描的点云数据处理

因具备高速、灵活和高精度的特点，移动式激光扫描被广泛用于地铁隧道的监测系统中。针对现有数据处理方法的里程配准误差大、数据利用率低的问题，本文提出了从扫描到后处理的一体化数据转换方法。在预扫描阶段，对隧道进行预标定，根据速度曲线的概率密度确定噪声界限；在正式扫描阶段，标定小车匀速运动的开始计速点，仅在惯导速度超限的情况下更新里程；在后处理阶段，首次基于激光点云数据生成360°全景图用于病害监测，提高了用户交互性。试验结果表明，本文方法在50 m内的测量误差小于1.2 mm，优于已有的螺旋扫描方法。因此，本文方法更适用于传感器精度低，测量频率高，且监测隧道较长的移动式激光扫描系统。同时，生成的全景图为隧道病害监测提供新的发展方向。

Point cloud data processing based on helical laser scanning system

Due to the advantages of high speed, flexibility and high resolution, helical laser scanners have been widely used in subway tunnel detection systems. However, existing data processing schemes have large synchronization error and low rate of data utilization. We propose a method to improve the accuracy of localization and make full use of the point cloud data. The starting point of the uniform motion of the car is calibrated, and the navigation speed is updated only when exceeding the upper limit, which is determined by the noise of the tunnel road. Meanwhile, the 360° panorama used for disease monitoring is firstly generated from point cloud data to improve the user interaction. Results show that the positioning error is 4.2 mm in 50 m, which is superior to the existing schemes. The generated panorama provides a new developing direction for disease monitoring in subway tunnels.